Photomask defect identification method

Abstract

The invention discloses a photomask defect identification method, and belongs to the technical field of photomask detection. The method comprises the following steps: S1, acquiring a surface defect image; s2, preprocessing the surface defect image based on the imaging characteristics of the defect; s3, performing filtering algorithm correction on the preprocessed surface defect image; s4, performing colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain a particle region; s5, on the basis of morphological characteristics of defects, segmenting the particle region to obtain defect islands; and S6, performing morphological analysis on the defective island to obtain a defect type and a defect size. The photomask defect identification method provided by the invention combines morphological and colorimetric analysis methods to automatically process the surface defect image and extract the defect island, improves the accuracy of defect identification, gets rid of the constraint of the difference of different equipment imaging systems, realizes automatic batch processing, and saves the manpower and material investment in the mass production link.

Description

Technical Field

[0001] The present application relates to the field of semiconductor detection technology, and in particular to a mask defect recognition method. Background Art

[0002] In the semiconductor manufacturing process, the mask is a key component in the photolithography technology, which can transfer patterns such as circuits to the semiconductor substrate. Its surface quality will directly affect the yield and performance of the chip. Therefore, in the production and processing of the mask, the detection and control of the surface defects of the mask has become an important part of the semiconductor manufacturing process. With the improvement of the mask generation technology node, the control of the surface defects of the mask has also been raised to a new level. In the production process of the mask, the staff obtains the morphology and position of the surface defects through repeated surface inspections, and then reversely diagnoses the root cause of the problem. The metal residue and missing on the surface of the mask after etching are often traced back to the surface of the raw materials at the input end or the particles inside the equipment.

[0003] At present, the methods for detecting the surface of photomasks include manual visual detection and automated detection based on image detection software. Among them, the manual detection method often uses optical equipment to scan and shape the surface of the photomask, and manually measures the surface particles and defect sizes through the measuring tools of the equipment's own software. If the inspection results cannot be confirmed on site, the defect image will be retained in place, and the defects will be measured later through measuring tools such as rulers to determine whether they can be used for production. However, due to the speed of manual review, a large number of photomask raw materials are retained in place, which greatly wastes human resources and machine time, and the accuracy of manual review is difficult to guarantee.

[0004] Automated detection methods generally use image detection software for automatic judgment, but there are currently few image processing algorithms corresponding to photomasks, and when processing optical images, threshold segmentation or image segmentation methods derived from threshold segmentation are often used. Since the photomask raw material includes a photoresist layer, it seriously interferes with the regular propagation of light, resulting in a large amount of noise and threshold overlap in the imaging, making it difficult to achieve the expected detection effect. Therefore, there is an urgent need for a photomask defect recognition method that can effectively detect and classify defects on the photomask surface. Summary of the invention

[0005] In view of the shortcomings of the prior art mentioned above, the purpose of this application is to provide a mask defect identification method to solve the problems of low efficiency and difficulty in ensuring accuracy in the prior art, and to effectively detect mask defects with the characteristics of high efficiency and high accuracy.

[0006] In order to achieve the above-mentioned purpose and other related purposes, the present application provides a method for identifying mask defects, comprising the following steps:

[0007] Acquire surface defect images;

[0008] Preprocessing the surface defect image based on the imaging characteristics of the defect;

[0009] Performing filtering algorithm correction on the preprocessed surface defect image;

[0010] Performing colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain a particle area;

[0011] Based on the morphological characteristics of the defects, the particle region is segmented to obtain a number of defect islands;

[0012] The defect islands are subjected to morphological analysis to obtain defect size and defect type.

[0013] Optionally, acquiring the surface defect image comprises the following steps, including the following steps:

[0014] Acquire a mask surface image;

[0015] Setting a sample window for framing the surface defect image from the mask surface image;

[0016] The surface defect image is extracted from the mask surface image using the sample window.

[0017] Optionally, preprocessing the surface defect image comprises the following steps:

[0018] Setting a first mask for preprocessing, wherein the first mask has a center point e;

[0019] Calculating a first pixel threshold according to the grayscale value of a pixel point in the surface defect image, and extracting a first target point from the surface defect image according to the first pixel threshold;

[0020] Align the center point e with the first target point, and assign grayscale values ​​to the pixels of the first mask;

[0021] Setting a target grayscale interval, and extracting a second target point from the first mask according to the target grayscale interval;

[0022] Calculating a first grayscale mean value of a plurality of the second target points and assigning the first grayscale mean value to the first target point;

[0023] The assignment of values ​​to all first target points in the surface defect image is completed to complete the preprocessing of the surface defect image.

[0024] Optionally, the first mask is set to a 3×3 array of dots, and the center point e is located at the center of the first mask dot array;

[0025] The grayscale value of the first target point is assigned to the center point e, and the grayscale values ​​of the eight pixel points surrounding and adjacent to the first target point in the surface defect image are respectively assigned to the remaining eight pixel points in the first mask except the center point e, so as to complete the grayscale assignment of the first mask.

[0026] Optionally, extracting a first target point from the surface defect image according to the first pixel threshold comprises the following steps:

[0027] Obtaining the maximum grayscale value of the pixel points in the surface defect image;

[0028] Taking 1 / 3 of the maximum grayscale value of the surface defect image as the first pixel threshold;

[0029] A pixel point whose grayscale value is greater than the first pixel threshold is selected from the surface defect image as a first target point.

[0030] Optionally, extracting a second target point from the first mask according to the target grayscale interval comprises the following steps:

[0031] Obtaining the minimum grayscale value of pixels in the surface defect image;

[0032] Taking 2 / 4 of the minimum grayscale value of the surface defect image as the minimum value of the target grayscale interval, and taking 5 / 6 of the maximum grayscale value of the surface defect image as the maximum value of the target grayscale interval;

[0033] A pixel point whose grayscale value does not exceed the target grayscale interval is selected from the first mask as a second target point.

[0034] Optionally, performing filtering algorithm correction on the preprocessed surface defect image comprises the following steps:

[0035] Setting mean filter parameters, minimum filter parameters, a second mask for mean filtering, and a third mask for minimum filtering;

[0036] Based on the second mask and the mean filter parameters, performing mean shift filtering on the preprocessed surface defect image;

[0037] Based on the third mask and the minimum filtering parameters, minimum filtering is performed on the surface defect image after the mean shift filtering.

[0038] Optionally, performing colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain the particle area comprises the following steps:

[0039] The difference between the maximum grayscale value and the minimum grayscale value of the pixel point of the surface defect image is calculated and recorded as a first grayscale difference;

[0040] Extracting pixel points whose grayscale values ​​are less than or equal to 2 / 3 times of the first grayscale difference from the surface defect image as third target points;

[0041] Calculate the grayscale average value of the plurality of third target points and record it as the second grayscale average value;

[0042] The difference between the grayscale value of each pixel in the surface defect image and the second grayscale mean is calculated and recorded as a second grayscale difference;

[0043] Pixel points whose second grayscale difference is greater than the second pixel threshold are selected from the surface defect image to obtain a particle area.

[0044] Optionally, segmenting the particle region to obtain a plurality of defect islands comprises the following steps:

[0045] Performing Manhattan distance transformation on the particle region;

[0046] Based on the particle area after Manhattan distance transformation, the third pixel threshold is calculated;

[0047] According to the third pixel threshold, the particle region after Manhattan distance transformation is segmented by using a watershed algorithm to obtain a plurality of defect islands.

[0048] Optionally, performing morphological analysis on the defect island to obtain defect size and defect type comprises the following steps:

[0049] Obtain the rectangular coordinates of each pixel point in the defect island as pixel point coordinates;

[0050] Calculating the width and height of the defect island according to the pixel coordinates;

[0051] Calculating the size according to the width and height of the defect island and the magnification ratio of the surface defect image;

[0052] The grayscale average value of the defect island pixel point is calculated and recorded as the third grayscale average value;

[0053] According to the maximum value of the pixel coordinates of the defect island, a first endpoint h1, a second endpoint h2, a third endpoint h3 and a fourth endpoint h4 of the defect island are obtained;

[0054] Calculate the midpoint h 12 , midpoint h 34 and the coordinates of the center point O, and the expression of the center line through the center point O; wherein the midpoint h12 is the midpoint between the first endpoint h1 and the second endpoint h2. 34 is the midpoint between the third endpoint h3 and the fourth endpoint h4, and the center point O is the midpoint h 12 With the midpoint h 34 The midpoint of the center line is perpendicular to the line segment h 12 h 34 ;

[0055] According to the expression of the center line, a first boundary point b1 and a second boundary point b2 of the defect island are obtained;

[0056] Calculate the line segment b1b2 and line segment h 12 h 34 The ratio between the short side and the long side in is recorded as the aspect ratio;

[0057] According to the preset fourth pixel threshold and the depression ratio threshold, the defect type is determined by the third grayscale mean and the aspect ratio.

[0058] The mask defect recognition method provided by the present application has at least the following beneficial effects:

[0059] In the mask defect recognition method of the present application, a preprocessing process is used to effectively filter out background noise in the surface defect image, retain and enhance the characteristics of defect particles, enhance the contrast between defects and background, facilitate subsequent image processing, and ensure the accuracy of defect recognition; mean shift filtering is used in combination with minimum filtering to improve the filtering effect of noise in the surface defect image, and at the same time improve the enhancement effect of the chromaticity and morphological characteristics of the defects and their boundaries; by calculating the third pixel threshold as the segmentation threshold of the watershed algorithm, effective segmentation of the defect area is achieved, the independence of the defect islands obtained by cutting is guaranteed, and the morphological and chromatic characteristics of the defects can be reflected, the defect islands are calculated to obtain the defect size, the third grayscale mean and the aspect ratio, the defect type can be effectively judged, the defect type can be automatically identified, and the detection efficiency and the accuracy of the detection results are improved; and when the sample image changes, the mask defect recognition method of this embodiment can be adaptively adjusted to get rid of the influence of equipment differences on the detection results, and can realize automatic detection and identification of defect size and type. The mask defect recognition method provided by the present invention combines morphological and colorimetric analysis methods to automatically process surface defect images to extract defect islands, thereby improving the accuracy of defect recognition, breaking free from the constraints of differences in imaging systems of different devices, realizing automatic batch processing, and saving manpower and material resources in the mass production link, which is extremely practical and worth promoting. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 Shown is a flow chart of the mask defect recognition method provided in the first embodiment of the present application.

[0062] Figure 2 It is a schematic diagram of the structure of over-cutting of an image in the prior art.

[0063] Figure 3 Shown is a schematic diagram of the process of preprocessing a surface defect image provided in Example 1 of the present application.

[0064] Figure 4 Shown is a schematic diagram comparing surface defect images before and after pretreatment provided in Example 1 of the present application.

[0065] Figure 5 Shown is a schematic diagram comparing surface defect images before and after filtering processing provided in Example 1 of the present application.

[0066] Figure 6 Shown is a schematic diagram of the process of obtaining a particle area from a surface defect image provided in Example 1 of the present application.

[0067] Figure 7 Shown is a schematic diagram of the process of obtaining particle size and defect type provided in Example 1 of the present application.

[0068] Figure 8 Shown is a structural schematic diagram of a surface defect image with three defects provided in an optional embodiment of the first embodiment of the present application.

[0069] Fig. 9 A schematic diagram showing the positions of the first to fourth endpoints provided as an optional embodiment in Embodiment 1 of the present application.

[0070] Fig.10 A schematic diagram showing the positions of the first to fourth endpoints provided as another optional embodiment in the first embodiment of the present application. DETAILED DESCRIPTION

[0071] In order to make the technical purpose, technical solution and technical effect of the present application clearer, the technical solution in the present application will be clearly and completely described in conjunction with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0072] Therefore, the following detailed description of the embodiments of the present application is not intended to limit the scope of the present application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present application. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0073] In the description of the present application, it should be noted that the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "example", "specific example" or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the implementation or example are included in at least one implementation or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same implementation or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more implementations or examples in a suitable manner.

[0074] This embodiment provides a method for identifying photomask defects, which is used to solve the problems of low detection efficiency and difficulty in ensuring accuracy of photomask defect detection and identification in the prior art. Figure 1 The mask defect recognition method provided in this embodiment includes steps S1 to S6.

[0075] Step S1: Acquire a surface defect image.

[0076] In this embodiment, the surface defect image is used to provide a sample for subsequent defect identification. Optionally, obtaining the surface defect image includes the following steps: obtaining the mask surface image; setting a sample window for taking the surface defect image from the mask surface image frame; and extracting the surface defect image from the mask surface image using the sample window.

[0077] Specifically, an optical device can be used to obtain a mask surface image, thereby providing a data sample for the mask defect recognition method of this embodiment; after obtaining the mask surface image, the size and shape of the sample window are set, and the sample window is used as a selection box; the sample window is used to frame the corresponding surface defect image from the sample image; wherein the surface defect image has one or more particles, and the particles are images of the mask defects in the surface defect image.

[0078] In an optional embodiment, the mask surface image is, for example, a rectangular image of 480×640 pixels, and the mask surface image contains defect information, position information, equipment information, etc. The sample window is set to a rectangular window of 200×300 pixels, and the sample window is used to frame a surface defect image of 200×300 pixels from the mask surface image. The size of 200×300 pixels corresponds to a field of view size of 12.5μm×18.75μm, which can frame sufficient defect information to provide sufficient data samples for defect identification of the mask, and can also remove other information irrelevant to defect identification, such as position information and equipment information. Framed surface defect images from the mask surface image can be set as an adaptive process. Step S1 is a necessary step before preprocessing in the mask defect identification method provided in this embodiment, which can effectively reduce noise interference and data processing volume, and improve detection efficiency.

[0079] Step S2: pre-processing the surface defect image based on the imaging characteristics of the defect.

[0080] There is a lot of noise in the surface defect image. In the prior art, Figure 2 In the process of image processing, image segmentation methods are often used to further process the image. Generally, image segmentation methods are extremely sensitive to the presence of noise. When there is a lot of noise, it will seriously affect the image segmentation quality. There is a large amount of overlap between the defect grayscale value and the background grayscale value in the image. Multiple thresholds are needed to segment multiple defects, which is prone to over-segmentation and other undesirable phenomena.

[0081] Based on the above problems, in this embodiment, the surface defect image is preprocessed to filter the background noise in the surface defect image to prevent the over-segmentation phenomenon. Specifically, refer to Figure 3 , preprocessing the surface defect image, including steps S21 to S26.

[0082] S21: Setting a first mask for preprocessing; wherein the first mask has a center point e and a plurality of pixel points surrounding the center point e. Optionally, the first mask may be, for example, a mask of a 3×3 array of dots, including 9 points, namely, the center point e, and points e1, e2, e3, e4, e5, e6, e7, and e8 surrounding the center point e.

[0083] S22: Calculate a first pixel threshold according to the grayscale value of the pixel points in the surface defect image, and extract a first target point from the surface defect image according to the first pixel threshold. Specifically, obtain the maximum grayscale value of the pixel points in the surface defect image, use 1 / 3 of the maximum grayscale value as the first pixel threshold, and select a pixel point with a grayscale value greater than the first pixel threshold from the surface defect image as the first target point.

[0084] In this embodiment, setting the first pixel threshold to 1 / 3 of the maximum grayscale value of the surface defect image can achieve the best preprocessing effect, retain and enhance the characteristics of the defect, increase the contrast between the defect and the background, and effectively filter the background noise in the surface defect image, which can maximize the effect of subsequent image processing and ensure the effective identification of the type and size of the mask defect; and when the surface defect image changes, the first pixel threshold can adaptively change with the surface defect image to ensure that the preprocessing effect is not affected, thereby getting rid of the influence of differences in optical equipment types on the mask defect recognition and detection effect.

[0085] S23: Align the center point e of the first mask with the first target point, and assign grayscale values ​​to the pixels of the first mask. Specifically, the grayscale value of the first target point is assigned to the center point e, and the grayscale values ​​of the pixels surrounding and adjacent to the first target point in the surface defect image are respectively assigned to the other pixels in the first mask except the center point e, so that the pixels surrounding the center point e correspond one-to-one with the pixels surrounding the first target point, thereby completing the assignment of the pixels of the first mask.

[0086] In an optional embodiment, the first mask is a 3×3 mask, including 9 points, namely a center point e, and points e1, e2, e3, e4, e5, e6, e7 and e8 surrounding the center point e, and 8 pixel points surrounding the first target point and adjacent to the first target point are used to perform grayscale assignment on points e1 to e8 respectively to complete the assignment of pixel points of the first mask.

[0087] S24: Set a target grayscale interval, and extract a second target point from the first mask according to the target grayscale interval. Specifically, respectively extract the grayscale values ​​of multiple pixels in the first mask, obtain the minimum grayscale value of the pixels in the surface defect image, and determine whether the grayscale value of the pixel of the first mask exceeds the range of the target grayscale interval. If it does not exceed the range of the target grayscale interval, use the corresponding pixel as the second target point.

[0088] In an optional embodiment, the first mask is a 3×3 mask, and whether the grayscale values ​​of the center point e, point e1, point e2, point e3, point e4, point e5, point e6, point e7 and point e8 exceed the target grayscale range is determined respectively. When the grayscale values ​​of the above pixel points are within the target grayscale range (including the endpoints), the corresponding pixel points are used as the second target points.

[0089] Furthermore, 2 / 4 of the minimum grayscale value of the surface defect image is used as the minimum value of the target grayscale interval, and 5 / 6 of the maximum grayscale value of the surface defect image is used as the maximum value of the target grayscale interval to form the target grayscale interval. Setting the endpoint value of the target grayscale interval to the above value can maximize the retention of the features of the first target point while improving the contrast between the defect and the background. Moreover, when the surface defect image changes, the target grayscale interval can adaptively change with the surface defect image, thus getting rid of the influence of differences in imaging systems of different devices.

[0090] S25: Calculate the first grayscale mean of the plurality of second target points and assign it to the first target point.

[0091] Specifically, the grayscale average value of the plurality of second target points is calculated and recorded as the first grayscale average value, and the first grayscale average value is assigned to the first target point.

[0092] S26: Complete the assignment of all first target points in the surface defect image to complete the preprocessing of the surface defect image. Specifically, perform steps S23 to S25 on the first target points in the surface defect image that have not undergone the above operations to complete the grayscale assignment of all first target points.

[0093] In an optional embodiment, steps S21 to S26 may be iterated to improve the preprocessing effect on the surface defect image. Optionally, the number of iterations is 1 or 2, at which time the best preprocessing effect can be achieved.

[0094] In this embodiment, refer to Figure 4 By setting the first pixel threshold and the target grayscale interval to pre-process the surface defect image, the background noise in the surface defect image is effectively eliminated, the contrast between the defect and the background is increased, and the occurrence of over-segmentation is prevented. The accuracy and reliability of subsequent data processing are greatly improved, thereby ensuring the effect of mask defect detection and identification, and getting rid of the influence of surface defect image differences caused by differences in imaging system types or parameters of different equipment on the detection results. It can realize automatic detection of sample images, improve detection efficiency, save labor costs, and improve the accuracy and reliability of detection results.

[0095] Step S3: Correct the preprocessed surface defect image using a filtering algorithm.

[0096] In this embodiment, a suitable filtering method is selected to perform filtering algorithm correction on the preprocessed surface defect image to remove noise in the surface defect image. Optionally, mean shift filtering and minimum filtering are used to filter the surface defect image. Specifically, the filtering algorithm correction is performed on the preprocessed surface defect image, including steps S31 to S33.

[0097] S31: Set mean filter parameters, minimum filter parameters, a second mask plate for mean filtering, and a third mask plate for minimum filtering. The second mask plate is used for mean migration filtering, and the mean filter parameters include its physical space radius, chrominance space radius, mean filter iteration number, and migration stop condition; the third mask plate is used for minimum filtering, and the minimum filter parameters include the iteration number of minimum filtering.

[0098] In an optional embodiment, the second mask is a mask of a 3×3 array dot matrix, the physical space radius of the mean filter parameter is 3, the chromaticity space radius is 20, the number of mean filter iterations is 5, and the migration stop condition is that the number of iterations reaches the number of mean filter iterations or the grayscale value of the center point of the second mask no longer changes. The third mask is set to a 3×3 mask, and the number of iterations in the minimum filter parameter is 1.

[0099] S32: Based on the second mask and the mean filter parameters, perform mean shift filtering on the preprocessed surface defect image.

[0100] Specifically, the center point of the second mask is aligned with the fourth target point in the surface defect image, and the grayscale value of the center point of the second mask is equal to the grayscale value of the fourth target point. The pixels surrounding the center point of the second mask are aligned one by one with the pixels surrounding the fourth target point, and the corresponding grayscale values ​​are assigned to the pixels of the second mask, so that the center point of the second mask traverses all the pixels of the surface defect image, thereby realizing the mean migration filtering of the surface defect image. When the center point of the second mask is migrated, the positions of the pixels surrounding the center point change accordingly, and when the migration stop condition is met, the migration stops.

[0101] S33: Based on the third mask and the minimum filtering parameters, perform minimum filtering on the surface defect image after the mean shift filtering. Specifically, similar to step S32, align the center point of the third mask with the fifth target point of the surface defect image, complete the grayscale assignment of the third mask, assign the minimum pixel value in the third mask to the fifth target point, and complete the minimum filtering according to the minimum filtering parameters.

[0102] In this embodiment, refer to Figure 5, the noise in the surface defect image can be further filtered by cooperating with the mean shift filter and the minimum filter, and the chromaticity characteristics and morphological characteristics of the defects and their boundaries in the surface defect image can be enhanced to make up for the missing defect boundaries; in addition, the second mask and the third mask are both set to 3×3 masks, the physical space radius is set to 3, the chromaticity space radius is set to 20, and the number of iterations is set to 5 and 1 respectively, which can optimize the filtering and enhancement effects; and in this embodiment, when other methods are used for filtering or the relevant parameters during filtering are changed, the filtering effect will be significantly reduced, and the enhancement effect on the defects and their boundary characteristics will not be as expected, thereby affecting the accuracy and precision of subsequent optical defect recognition detection results.

[0103] Step S4: Perform colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain the particle area.

[0104] In this embodiment, according to the preset second pixel threshold, the surface defect image after filtering is subjected to colorimetric analysis and recognition, and then the image with defective particles is extracted from the surface defect image and recorded as the particle area. By extracting the particle area, the more obvious background area is removed, which can facilitate subsequent data processing, reduce the amount of calculation, and improve the calculation efficiency. Optionally, refer to Figure 6 , performing colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain the particle area includes steps S41 to S45.

[0105] S41: Calculate the difference between the maximum grayscale value and the minimum grayscale value of the pixel point in the surface defect image and record it as the first grayscale difference. Specifically, obtain the maximum grayscale value and the minimum grayscale value of the pixel point in the filtered surface defect image, and subtract the minimum grayscale value from the maximum grayscale value to obtain the first grayscale difference.

[0106] S42: extracting a third target point from the surface defect image according to the first grayscale difference. Optionally, extracting a pixel point whose grayscale value is less than or equal to 2 / 3 times the first grayscale difference from the surface defect image as the third target point.

[0107] S43: Calculate the grayscale average value of the plurality of third target points and record it as the second grayscale average value.

[0108] S44: The difference between the grayscale value of each pixel in the surface defect image and the second grayscale mean is calculated and recorded as the second grayscale difference.

[0109] S45: Select pixel points whose second grayscale difference is greater than the second pixel threshold from the surface defect image to obtain the particle area. Specifically, determine whether the second grayscale difference is greater than the preset second pixel threshold. When the second grayscale difference is greater than the second pixel threshold, determine that the corresponding pixel point is a pixel point in the particle area. When the second grayscale difference is less than or equal to the second pixel threshold, determine that the corresponding pixel point is a pixel point in the background area. Obtain all pixel points whose second grayscale difference is greater than the second pixel threshold in the detection pattern to form a particle area. Optionally, the second pixel threshold is set to 50.

[0110] In this embodiment, by setting 2 / 3 of the first grayscale difference to select the third target point, setting the second pixel threshold to 50, and using the second grayscale difference to obtain the particle area from the surface defect image, the obviously irrelevant background area can be effectively removed, and the best recognition effect can be obtained. At the same time, it is ensured that the particle area contains the defects in the surface defect image, reducing the amount of data processing calculations and improving the efficiency of mask defect detection. In addition, when the surface defect image changes, the first grayscale difference and the second grayscale difference can change adaptively with the surface defect image, thereby getting rid of the influence of the imaging system types or parameter differences of different devices, and realizing automatic recognition of the particle area.

[0111] Step S5: Based on the morphological characteristics of the defects, the particle region is segmented to obtain a number of defect islands.

[0112] In this embodiment, there are one or more defective particles in the particle region. Different defective particles may be independent of each other, or may overlap with each other to form a defective particle with a larger size. By segmenting the particle region using an image segmentation method, one or more defective particles independent of each other can be obtained and recorded as defect islands to reflect the morphological characteristics of independent defects.

[0113] In an optional embodiment, segmenting the particle region to obtain a plurality of defect islands includes the following steps: performing Manhattan distance transformation on the particle region; calculating a third pixel threshold based on the particle region after the Manhattan distance transformation; segmenting the particle region after the Manhattan distance transformation using a watershed algorithm according to the third pixel threshold to obtain one or more independent defect islands; wherein a defect island is a set of defect pixel points that can reflect the morphological and chromatic characteristics of the defect; optionally, segmenting the particle region to obtain defect islands also includes the following steps: obtaining the number of defect islands according to the obtained defect islands.

[0114] After the Manhattan distance transformation, for each pixel point in the particle area, the farther the distance from the defect boundary is, the higher the gray value of the pixel point is, and the closer the distance from the defect boundary is, the lower the gray value of the pixel point is. Optionally, the third pixel threshold is 2 / 5 of the maximum gray value of the pixel point in the particle area after the Manhattan distance transformation, and the third pixel threshold is used as the threshold of the watershed algorithm to segment the particle area.

[0115] In this embodiment, by calculating the third pixel threshold and using it as the segmentation threshold of the watershed algorithm, an organic combination of chromaticity features and morphological features is achieved, which effectively ensures that the defect islands obtained by cutting are independent of each other and can reflect the morphological and chromatic characteristics of the actual defects, thereby achieving effective segmentation of the particle area; and when the surface defect image changes, the third pixel threshold changes accordingly, ensuring adaptive adjustment of the particle area segmentation process and realizing the automation of particle area segmentation.

[0116] Step S6: Perform morphological analysis on the defect island to obtain the defect size and defect type.

[0117] Specifically, in this embodiment, the size of the defect island is calculated as the defect size, and the defect type is determined based on the shape of the defect island as the defect type. Figure 7 , data processing is performed on the defect island to obtain the defect size and defect type, including steps S61 to S69.

[0118] S61: Obtain the rectangular coordinates of each pixel point in the defect island and mark them as pixel point coordinates.

[0119] In this embodiment, the rectangular coordinates of each pixel point of the defect island include a horizontal coordinate and a vertical coordinate. Optionally, a pixel point can be selected from the defect island as the coordinate origin, and two mutually perpendicular directions can be selected as the x-axis and the y-axis respectively to establish a rectangular coordinate system, and then the rectangular coordinates of each pixel point of the defect island are obtained; or, a rectangular coordinate system is pre-set in the surface defect image, and the coordinates of each pixel point in the defect island are obtained according to the rectangular coordinate system.

[0120] S62: Obtain the width and height of the defect island according to the pixel coordinates.

[0121] Specifically, the maximum horizontal coordinate, minimum horizontal coordinate, maximum vertical coordinate and minimum vertical coordinate of the pixel point in the defect island are obtained, and the difference between the maximum horizontal coordinate and the minimum horizontal coordinate is calculated as the width of the defect island, and the difference between the maximum vertical coordinate and the minimum vertical coordinate is calculated as the height of the defect island.

[0122] S63: Calculate the defect size according to the width and height of the defect island and the magnification ratio of the surface defect image.

[0123] Specifically, the ratio between the width of the defect island and the magnification ratio of the surface defect image is calculated to obtain the width of the defect; and the ratio between the height of the defect island and the magnification ratio of the surface defect image is calculated to obtain the height of the defect.

[0124] In an optional embodiment, for example, a sample image of 480×640 pixels corresponds to a viewing area size of 30 μm×40 μm. The width and height of the defect at the same magnification ratio can be calculated using the above relationship. Figure 8 For example, the sizes of defect islands are 3×4 pixels, 7×7 pixels, and 6×6 pixels, respectively. The corresponding viewing area sizes are calculated to be 0.1875μm×0.25μm, 0.4375μm×0.4375μm, and 0.375μm×0.375μm.

[0125] S64: Calculate and record the grayscale average value of the defect island pixels as the third grayscale average value.

[0126] S65: Obtain a first endpoint h1, a second endpoint h2, a third endpoint h3 and a fourth endpoint h4 of the defect island according to the maximum value of the pixel coordinates of the defect island.

[0127] In an alternative embodiment, reference Fig. 9 , according to the maximum value of the ordinate of the pixel point in the defect island, the first endpoint h1, the second endpoint h2, the third endpoint h3 and the fourth endpoint h4 of the defect island are obtained. Specifically, the maximum value and the minimum value of the ordinate of the pixel point in the defect island are obtained; when the ordinate takes the maximum value, the value of the horizontal coordinate has a minimum value and a maximum value, and the maximum value of the ordinate and the corresponding maximum value of the horizontal coordinate are used as the ordinate and the horizontal coordinate of the first endpoint h1, respectively, and the maximum value of the ordinate and the corresponding minimum value of the horizontal coordinate are used as the ordinate and the horizontal coordinate of the second endpoint h2, respectively; when the ordinate takes the minimum value, the value of the horizontal coordinate has a maximum value and a minimum value, and the minimum value of the ordinate and the corresponding maximum value of the horizontal coordinate are used as the ordinate and the horizontal coordinate of the third endpoint h3, respectively, and the minimum value of the ordinate and the corresponding minimum value of the horizontal coordinate are used as the ordinate and the horizontal coordinate of the fourth endpoint h4, respectively.

[0128] In another alternative embodiment, referring to Fig.10, according to the maximum value of the horizontal coordinate of the pixel point in the defect island, the first endpoint h1, the second endpoint h2, the third endpoint h3 and the fourth endpoint h4 of the defect island are obtained. Specifically, the maximum value and the minimum value of the horizontal coordinate of the pixel point in the defect island are obtained; when the horizontal coordinate takes the maximum value, the vertical coordinate has a minimum value and a maximum value, and the maximum value of the horizontal coordinate and the corresponding maximum value of the vertical coordinate are used as the horizontal coordinate and the vertical coordinate of the first endpoint h1, respectively, and the maximum value of the horizontal coordinate and the corresponding minimum value of the vertical coordinate are used as the horizontal coordinate and the vertical coordinate of the second endpoint h2, respectively; when the horizontal coordinate takes the minimum value, the vertical coordinate has a maximum value and a minimum value, and the minimum value of the horizontal coordinate and the corresponding maximum value of the vertical coordinate are used as the horizontal coordinate and the vertical coordinate of the third endpoint h3, respectively, and the minimum value of the horizontal coordinate and the corresponding minimum value of the vertical coordinate are used as the horizontal coordinate and the vertical coordinate of the fourth endpoint h4, respectively.

[0129] S66: Calculate the midpoint h 12 , midpoint h 34 and the coordinates of the center point O, and the expression of the center line through the center point O. Among them, the midpoint h 12 is the midpoint between the first endpoint h1 and the second endpoint h2. 34 The midpoint of the third endpoint h3 and the fourth endpoint h4, the center point O is the midpoint h 12 With midpoint h 34 The midpoint of the line is perpendicular to the line segment h 12 h 34 , line segment h 12 h 34 The midpoint h 12 With midpoint h 34 connection.

[0130] In this embodiment, the coordinates of the first endpoint h1 are (x1, y1), the coordinates of the second endpoint h2 are (x2, y2), the coordinates of the third endpoint h3 are (x3, y3), the coordinates of the fourth endpoint h4 are (x4, y4), and the midpoint h 12 The coordinates of (x 12 ,y 12 ), midpoint h 34 The coordinates of (x 34 ,y 34 ), the coordinates of the center point O are (x0, y0), then x 12 =0.5(x1+x2),y 12 =0.5(y1+y2),x 34 =0.5(x3+x4),y 34 =0.5(y3+y4), x0=0.5(x 12 +x 34 ), y0=0.5(y 12 +y 34). When x1+x2 is an odd number, 1 is automatically added. Similarly, the above operation is applied to the remaining coordinates.

[0131] S67: According to the center line expression, obtain the first boundary point b1 and the second boundary point b2 of the defect island.

[0132] In this embodiment, pixel points are iterated cyclically along the center line toward both sides of the center point O, and it is determined whether the current pixel point of the cyclic iteration is located in the defect island, thereby obtaining the first boundary point b1 and the second boundary point b2 of the defect island.

[0133] In an optional embodiment, the expression of the center line is y=kx+b, and the coordinates of the pixel points iterated along the center line to both sides of the center point O can be expressed as (x0±x n+1 , y0±y n+1 ), where (x n+1 ,y n+1 ) is the coordinate of the pixel point on the center line, n is the number of the pixel points along the center line that are away from the center point O in sequence, and the value is an integer greater than or equal to 0. Specifically, n is set to 0, and the value of n is increased in sequence, and the pixel point (x0±x n+1 , y0±y n+1 ) is located in the defect island. When the next pixel point is outside the defect island, the current pixel point is used as the boundary point of the defect island, and then the first boundary point b1 and the second boundary point b2 are obtained.

[0134] S68: Calculate the line segment b1b2 and line segment h 12 h 34 The ratio between the short side and the long side in is recorded as the aspect ratio. Fig. 9 and Fig.10 , according to the midpoint h 12 and midpoint h 34 The coordinates of the line segment h are calculated 12 h 34 The length of the line segment b1b2 is calculated based on the coordinates of the first boundary point b1 and the second boundary point b2. 12 h 34 , calculate the ratio of the short side length to the long side length, and record it as the aspect ratio. The aspect ratio is less than or equal to 1.

[0135] S69: judging the defect type by the third grayscale mean value and the aspect ratio according to the preset fourth pixel threshold and the recess ratio threshold.

[0136] Specifically, in this embodiment, when the third grayscale mean is greater than the fourth pixel threshold, the defect type is determined to be a depression, when the third grayscale mean is less than or equal to the fourth pixel threshold, the defect type is determined to be a particle, when the third grayscale mean is greater than the third pixel threshold and the aspect ratio is less than the depression ratio threshold, the defect type is a crack, when the third grayscale mean is less than or equal to the fourth pixel threshold and the aspect ratio is greater than or equal to the depression ratio threshold, the defect type is a flocculent particle. Optionally, the fourth pixel threshold is set to 200, and the depression ratio threshold is set to 0.1.

[0137] In the mask defect recognition method provided in the present embodiment, the surface defect image is preprocessed, the background noise in the surface defect image is effectively filtered, the characteristics of the particles are retained and enhanced, the contrast between the particles and the background is enhanced, and the effect of subsequent image processing is improved; the mean shift filter is used in combination with the minimum filter, and specific parameters are set to improve the filtering effect of the noise in the surface defect image, while improving the enhancement effect of the chromaticity characteristics and morphological characteristics of the particles and their boundaries, and further enhancing the contrast; by calculating the first grayscale difference and according to the second pixel threshold, the particle area is obtained from the surface defect image, and the third pixel threshold is calculated as the segmentation threshold of the watershed algorithm, so that the effective segmentation of the particle area is achieved, the independence of the defect islands obtained by cutting is guaranteed, and the morphological characteristics and chromaticity characteristics of the defective particles can be reflected; by calculating the size of the defect island as the defect size output, according to the third grayscale mean and the aspect ratio of the defect island, the type of defect can be judged and output; and when the sample image changes, the mask defect recognition method of the present embodiment can be adaptively adjusted, get rid of the influence of the imaging system differences or parameter differences of different devices on the detection results, and realize the automation of the defect recognition process. Compared with the existing technology, the mask defect recognition method of this embodiment combines morphological and colorimetric analysis methods to automatically process surface defect images to extract defect islands, effectively improving the applicability, accuracy and reliability of the method, breaking away from the constraints of differences in imaging systems of different devices, and improving detection efficiency, saving manpower and material costs, and is extremely practical and worth promoting.

[0138] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify, change or combine the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A method for identifying mask defects, characterized in that: The following steps are involved: Acquire surface defect images; Preprocessing the surface defect image based on the imaging characteristics of the defect; Performing filtering algorithm correction on the preprocessed surface defect image; Performing colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain a particle area; Based on the morphological characteristics of the defects, the particle region is segmented to obtain a number of defect islands; The defect islands are subjected to morphological analysis to obtain defect size and defect type.

2. The method for identifying mask defects according to claim 1, characterized in that: Acquiring surface defect images includes the following steps: Acquire a mask surface image; Setting a sample window for framing the surface defect image from the mask surface image; The surface defect image is extracted from the mask surface image using the sample window.

3. The method for identifying mask defects according to claim 1, characterized in that: Preprocessing the surface defect image includes the following steps: Setting a first mask for preprocessing, wherein the first mask has a center point e; Calculating a first pixel threshold according to the grayscale value of a pixel point in the surface defect image, and extracting a first target point from the surface defect image according to the first pixel threshold; Align the center point e with the first target point, and assign grayscale values ​​to the pixels of the first mask; Setting a target grayscale interval, and extracting a second target point from the first mask according to the target grayscale interval; Calculating a first grayscale mean value of a plurality of the second target points and assigning the first grayscale mean value to the first target point; The assignment of values ​​to all first target points in the surface defect image is completed to complete the preprocessing of the surface defect image.

4. The method for identifying mask defects according to claim 3, characterized in that: The first mask is set to a 3×3 array of dots, and the center point e is located at the center of the first mask dot array; The grayscale value of the first target point is assigned to the center point e, and the grayscale values ​​of the eight pixel points surrounding and adjacent to the first target point in the surface defect image are respectively assigned to the remaining eight pixel points in the first mask except the center point e, so as to complete the grayscale assignment of the first mask.

5. The method for identifying mask defects according to claim 3, characterized in that: Extracting a first target point from the surface defect image according to the first pixel threshold comprises the following steps: Obtaining the maximum grayscale value of the pixel points in the surface defect image; Taking 1 / 3 of the maximum grayscale value of the surface defect image as the first pixel threshold; A pixel point whose grayscale value is greater than the first pixel threshold is selected from the surface defect image as a first target point.

6. The method for identifying mask defects according to claim 3 or 5, characterized in that: Extracting a second target point from the first mask according to the target grayscale interval includes the following steps: Obtaining the minimum grayscale value of pixels in the surface defect image; Taking 2 / 4 of the minimum grayscale value of the surface defect image as the minimum value of the target grayscale interval, and taking 5 / 6 of the maximum grayscale value of the surface defect image as the maximum value of the target grayscale interval; A pixel point whose grayscale value does not exceed the target grayscale interval is selected from the first mask as a second target point.

7. The method for identifying mask defects according to claim 1, characterized in that: The pre-processed surface defect image is corrected by a filtering algorithm, comprising the following steps: Setting mean filter parameters, minimum filter parameters, a second mask for mean filtering, and a third mask for minimum filtering; Based on the second mask and the mean filter parameters, performing mean shift filtering on the preprocessed surface defect image; Based on the third mask and the minimum filtering parameters, minimum filtering is performed on the surface defect image after the mean shift filtering.

8. The method for identifying mask defects according to claim 1, characterized in that: Performing colorimetric analysis on the surface defect image corrected by the filtering algorithm to obtain the particle area includes the following steps: The difference between the maximum grayscale value and the minimum grayscale value of the pixel point of the surface defect image is calculated and recorded as a first grayscale difference; Extracting pixel points whose grayscale values ​​are less than or equal to 2 / 3 times of the first grayscale difference from the surface defect image as third target points; Calculate the grayscale average value of the plurality of third target points and record it as the second grayscale average value; The difference between the grayscale value of each pixel in the surface defect image and the second grayscale mean is calculated and recorded as a second grayscale difference; Pixel points whose second grayscale difference is greater than the second pixel threshold are selected from the surface defect image to obtain a particle area.

9. The method for identifying mask defects according to claim 1, characterized in that: The particle region is segmented to obtain a plurality of defect islands, comprising the following steps: Performing Manhattan distance transformation on the particle region; Based on the particle area after Manhattan distance transformation, the third pixel threshold is calculated; According to the third pixel threshold, the particle region after Manhattan distance transformation is segmented by using a watershed algorithm to obtain a plurality of defect islands.

10. The method for identifying mask defects according to claim 1, characterized in that: Performing morphological analysis on the defect island to obtain defect size and defect type includes the following steps: Obtain the rectangular coordinates of each pixel point in the defect island as pixel point coordinates; Calculating the width and height of the defect island according to the pixel coordinates; Calculating the defect size according to the width and height of the defect island and the magnification ratio of the surface defect image; The grayscale average value of the defect island pixel point is calculated and recorded as the third grayscale average value; According to the maximum value of the pixel coordinates of the defect island, a first endpoint h1, a second endpoint h2, a third endpoint h3 and a fourth endpoint h4 of the defect island are obtained; Calculate the midpoint h 12 , midpoint h 34 and the coordinates of the center point O, and the expression of the center line through the center point O; wherein the midpoint h 12 is the midpoint between the first endpoint h1 and the second endpoint h2. 34 is the midpoint between the third endpoint h3 and the fourth endpoint h4, and the center point O is the midpoint h 12 With the midpoint h 34 The midpoint of the center line is perpendicular to the line segment h 12 h 34 ; According to the expression of the center line, a first boundary point b1 and a second boundary point b2 of the defect island are obtained; Calculate the line segment b1b2 and line segment h 12 h 34 The ratio between the short side and the long side in is recorded as the aspect ratio; According to the preset fourth pixel threshold and the depression ratio threshold, the defect type is determined by the third grayscale mean and the aspect ratio.

Images