Detection Method, Detection Device, Equipment and Storage Medium

Abstract

A detection method, detection device, equipment and storage medium. The detection method performs cutting channel and unit pattern filtering processing or cutting channel avoidance processing on the image to be processed according to the size of the unit image in the image to be processed, and can eliminate the interference caused by the cutting channel area to the detection result in the subsequent detection process of defective pixel points, so the accuracy of defective pixel point detection can be improved, and further the accuracy of defect detection can be improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular, to a detection method, a detection device, a device, and a storage medium. Background Art

[0002] With the continuous development of technology, precision machining is used in more and more fields. At the same time, there are also higher and higher requirements for machining accuracy.

[0003] In order to meet the requirements of machining accuracy and improve the qualified rate of products, it is necessary to perform on-line inspection on products. For example, by performing defect detection to determine whether there are defects in the products, and detecting the position and size of the defects to ensure that the relevant index requirements of product manufacturing are met.

[0004] However, the accuracy of the current detection results still needs to be improved. Summary of the Invention

[0005] The problem solved by the present invention is to provide a detection method, a detection device, a device, and a storage medium to improve the accuracy of detection.

[0006] To solve the above problems, the present invention provides a detection method, which includes:

[0007] Obtain an image to be processed; the image to be processed includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane region; the unit image includes a unit pattern;

[0008] Obtain the size of the unit image in the image to be processed;

[0009] Compare the size of the unit image in the image to be processed with a preset first size threshold;

[0010] If the size of the unit image is smaller than the first size threshold, perform scribe lane and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; perform first recognition processing on the first image to be detected, and obtain pixel points that meet the first threshold condition in the first image to be detected as defect pixel points;

[0011] If the size of the unit image is greater than or equal to the first size threshold, perform scribe lane avoidance processing on the image to be processed to obtain a second image to be detected; perform second recognition processing on the second image to be detected, and obtain pixel points that meet the second threshold condition in the second image to be detected as defect pixel points;

[0012] Based on the defect pixel points, obtain the defect points existing in the image to be processed.

[0013] Accordingly, an embodiment of the present invention further provides a detection device, the device comprising:

[0014] An image acquisition unit, adapted to acquire an image to be processed; the image to be processed includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane region; the unit image includes a unit pattern;

[0015] A size acquisition unit, adapted to acquire the size of the unit image in the image to be processed;

[0016] A size comparison unit, adapted to compare the size of the unit image in the image to be processed with a preset first size threshold;

[0017] An identification processing unit, adapted to, if the size of the unit image is less than the first size threshold, perform scribe lane and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; perform first identification processing on the first image to be detected, and obtain the pixel points satisfying the first threshold condition in the first image to be detected as defective pixel points; if the size of the unit image is greater than or equal to the first size threshold, perform scribe lane avoidance processing on the image to be processed to obtain a second image to be detected; perform second identification processing on the second image to be detected, and obtain the pixel points satisfying the second threshold condition in the second image to be detected as defective pixel points;

[0018] A defect acquisition unit, adapted to obtain the defect points existing in the image to be processed based on the defective pixel points.

[0019] Accordingly, an embodiment of the present invention further provides a device, comprising at least one memory and at least one processor, the memory storing one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the detection method according to any one of the above.

[0020] Accordingly, an embodiment of the present invention further provides a storage medium, the storage medium storing one or more computer instructions, the one or more computer instructions being used to implement the detection method according to any one of the above.

[0021] Compared with the prior art, the technical solution of the present invention has the following advantages:

[0022] The detection method provided by the embodiment of the present invention includes: obtaining an image to be processed; the image to be processed includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane region; obtaining the size of the unit images in the image to be processed; if the size of the unit image is less than a preset first size threshold, performing scribe lane and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; performing a first recognition process on the first image to be detected, and obtaining the pixel points that meet the first threshold condition in the first image to be detected as defective pixel points; if the size of the unit image is greater than or equal to the first size threshold, performing scribe lane avoidance processing on the image to be processed to obtain a second image to be detected; performing a second recognition process on the second image to be detected, and obtaining the pixel points that meet the second threshold condition in the second image to be detected as defective pixel points; based on the defective pixel points, obtaining the defective points existing in the image to be processed.

[0023] In the detection method of this embodiment, according to the size of the unit images in the obtained image to be processed, performing scribe lane and unit pattern filtering processing or scribe lane avoidance processing on the image to be processed can eliminate the interference caused by the scribe lane region to the detection result in the subsequent detection process of defective pixel points, so the accuracy of defective pixel point detection can be improved, and further the accuracy of defect detection can be improved. Description of the Drawings

[0024] Figure 1 is a flowchart of a detection method in an embodiment of the present invention;

[0025] Figure 2 is a partial schematic diagram of an image to be processed in an embodiment of the present invention;

[0026] Figure 3 is a schematic diagram of a unit image in the image to be processed in an embodiment of the present invention;

[0027] Figure 4 shows a schematic diagram of a spectral matrix obtained by performing Fourier transform on the image to be processed;

[0028] Figure 5 shows a schematic diagram of a first image to be detected in an embodiment of the present invention;

[0029] Figure 6 shows a schematic diagram of a positional relationship between a pixel point to be detected and a corresponding reference pixel point in an embodiment of the present invention;

[0030] Figure 7 shows a partial schematic diagram of an inverse binary image in an embodiment of the present invention;

[0031] Figure 8It shows a schematic diagram of a positional relationship between the image to be inspected and the corresponding reference image in an embodiment of the present invention;

[0032] Figure 9 It shows a schematic structural diagram of a detection device in an embodiment of the present invention;

[0033] Figure 10 It shows a hardware structure diagram of the device provided by an embodiment of the present invention. Detailed implementation manners

[0034] As can be seen from the background art, the accuracy of the current detection results still needs to be improved.

[0035] In the detection of the surface quality of products, image processing is usually used to identify the defects existing on the product surface. Specifically, the template matching algorithm is used to subtract the template image and the image to be processed to achieve the defect detection of the product.

[0036] The object to be measured includes a plurality of periodically arranged unit structures and scribe lines located between the unit structures. Correspondingly, the image to be processed includes a plurality of periodically arranged unit images and scribe line regions located between the unit images.

[0037] However, the scribe line region in the image to be processed is likely to interfere with the defect detection structure. Specifically, the pixel values between the pixel points in the scribe line region and the pixel points in the unit image are quite different, so that the pixel points in the scribe line region are likely to be misidentified as defect pixel points.

[0038] To solve the above problems, the present invention provides a detection method, including: obtaining an image to be processed; the image to be processed includes a plurality of unit images, and the plurality of unit images are separated from each other by scribe line regions; obtaining the size of the unit images in the image to be processed; if the size of the unit images is smaller than a preset first size threshold, performing scribe line and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; performing a first recognition process on the first image to be detected, and obtaining the pixel points that meet the first threshold condition in the first image to be detected as defect pixel points; if the size of the unit images is greater than or equal to the first size threshold, performing scribe line avoidance processing on the image to be processed to obtain a second image to be detected; performing a second recognition process on the second image to be detected, and obtaining the pixel points that meet the second threshold condition in the second image to be detected as defect pixel points; based on the defect pixel points, obtaining the defect points existing in the image to be processed.

[0039] In the detection method of this embodiment, according to the size of the unit images in the to-be-processed image obtained, the to-be-processed image is subjected to scribe lane and unit pattern filtering processing or scribe lane avoidance processing, which can eliminate the interference of the scribe lane area on the detection result in the subsequent detection process of defective pixel points. Therefore, the accuracy of defective pixel point detection can be improved, and further the accuracy of defect detection can be improved.

[0040] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings.

[0041] Figure 1 The flowchart of a detection method in an embodiment of the present invention is shown. Refer to Figure 1 The detection method can be implemented by the following steps:

[0042] Step S101: Obtain a to-be-processed image; the to-be-processed image includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane area; the unit image includes a unit pattern.

[0043] Step S102: Obtain the size of the unit image in the to-be-processed image.

[0044] Step S103: Compare the obtained size of the unit image with a preset first size threshold.

[0045] Step S104: If the size of the unit image is smaller than the preset first size threshold, perform scribe lane and unit pattern filtering processing on the to-be-processed image to obtain a first to-be-detected image; perform a first recognition process on the first to-be-detected image, and obtain the pixel points that meet the first threshold condition in the first to-be-detected image as defective pixel points.

[0046] Step S105: If the size of the unit image is greater than or equal to the first size threshold, perform scribe lane avoidance processing on the to-be-processed image to obtain a second to-be-detected image; perform a second recognition process on the second to-be-detected image, and obtain the pixel points that meet the second threshold condition in the second to-be-detected image as defective pixel points.

[0047] Step S106: Based on the defective pixel points, obtain the defective points existing in the to-be-processed image.

[0048] Please continue to refer to Figure 1 Perform step S101 to obtain a to-be-processed image; the to-be-processed image includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane area; the unit image includes a unit pattern (not shown).

[0049] Figure 2is a partial schematic diagram of an image to be processed; Figure 3 is a schematic diagram of a unit image in an image to be processed. Referring to Figure 2 and Figure 3 , the image 100 to be processed is an image that needs to be detected. In this embodiment, the image 100 to be processed is an image that needs to be defect-detected.

[0050] The image 100 to be processed is obtained by photographing an object to be measured. Specifically, the steps of obtaining the image 100 to be processed include: providing an imaging system and an object to be measured; using the imaging system to photograph the object to be measured to obtain the original image of the object to be measured; and obtaining the image to be processed based on the original image.

[0051] In this embodiment, the original image of the object to be measured has a certain degree of distortion. Therefore, after obtaining the original image of the object to be measured, the original image is corrected to eliminate the distortion generated by the original image, and the corrected original image is used as the image to be processed.

[0052] Specifically, the original image is subjected to a perspective transformation process to eliminate the distortion generated by the original image and obtain the image to be processed.

[0053] In other embodiments, when the original image of the object to be measured does not have distortion, the original image can also be directly used as the image to be processed.

[0054] A pixel is the smallest unit of an image. Therefore, the image 100 to be processed has a plurality of pixel points 101. Specifically, the plurality of pixel points 101 form a pixel array.

[0055] In this embodiment, the object to be measured includes a plurality of repeating unit structures and scribe lines. The plurality of repeating unit structures are separated from each other by the scribe lines, and a structural pattern is formed in each unit structure. Correspondingly, the image 100 to be processed correspondingly includes a plurality of identical unit images 110 and scribe line regions 120, and the unit images 110 have unit patterns. Among them, the unit image 110 is an image of the unit structure in the object to be measured, the scribe line region 120 is an image of the scribe line in the object to be measured, and the unit pattern is correspondingly an image of the structural pattern.

[0056] According to the arrangement of the plurality of repeating unit structures in the object to be measured, the unit images in the image to be processed are correspondingly arranged periodically. Figure 2 shows 12 unit images. Specifically, the 12 unit images 110 are arranged in a 4*3 array. It can be understood that the number of unit images 110 in the image 100 to be processed is not limited to 12.

[0057] It should be noted that for different objects to be measured, the sizes of the unit structures can be the same or different. Among them, if the size of the unit structure is large, the distributions of multiple periodically arranged unit structures and scribe lines in the object to be measured are relatively sparse. At this time, it is relatively difficult to perform filtering processing on the scribe lines. If the size of the unit structure is small, multiple periodically arranged unit structures and scribe lines are correspondingly more densely distributed in the object to be measured. At this time, the acquired image can be first subjected to Fourier low-pass filtering to obtain an object to be measured similar to an image-free object, and then a die defect detection method can be used to detect it.

[0058] In this embodiment, the object to be measured is a wafer. A wafer usually contains multiple repeated dies. Correspondingly, the image 100 to be processed is a wafer image, and each unit image 110 can include an image of one die or multiple dies.

[0059] In other embodiments, the object to be measured can also be other types of products such as a glass panel. It can be understood that the glass panel can also have multiple repeated unit structures. For example, each unit structure can be used to form an electronic product display screen.

[0060] Please continue to refer to Figure 1 , and perform step S102 to obtain the size of the unit image in the image to be processed.

[0061] Obtain the size of the unit image in the image to be processed, so as to compare the size of the unit image with a preset first size threshold in the subsequent process to obtain a corresponding comparison result.

[0062] Measure the unit image in the image to be processed to obtain the size of the unit image.

[0063] In this embodiment, the object to be measured is a wafer, and the unit image in the image to be processed is a die image. Correspondingly, the shape of the unit image is rectangular, and the unit image size includes length and width.

[0064] The image to be processed includes multiple pixel points, and the unit images in the image to be processed also respectively have corresponding multiple pixel points.

[0065] In this embodiment, the size of the unit image is measured by the number of pixel points. For example, the size of the unit image is 25*35. Specifically, the length of the unit image includes 25 pixel points, and the width of the unit image includes 35 pixel points. As Figure 3As shown, the length of the unit image is the dimension of the unit image along the first arrangement direction (X direction) of pixel points, and the width of the unit image is the dimension of the unit image along the second arrangement direction (Y direction) of pixel points; the first arrangement direction is perpendicular to the second arrangement direction.

[0066] In other embodiments, other methods can also be used to measure the size of the unit image, which can be selected by those skilled in the art according to actual needs and will not be limited herein.

[0067] Please continue to refer to Figure 1 , perform step S103 to compare the size of the obtained unit image with a preset first size threshold.

[0068] Subsequently, based on the result of the size of the unit image and the preset first size threshold, it is determined whether to perform cutting track avoidance processing or cutting track and unit pattern filtering processing on the image to be processed.

[0069] Here, comparing the size of the obtained unit image with the preset first size threshold means comparing the sizes of all unit images in the image to be processed with the first size threshold respectively.

[0070] In this embodiment, the size of the unit image includes the length and the width. Correspondingly, the first size threshold includes at least one of a length threshold and a width threshold.

[0071] The first size threshold can be set according to actual needs.

[0072] It can be understood that the first size threshold should not be too large or too small. When the first size threshold is too large, even if the size of the unit image is smaller than the first size threshold, the distribution of the unit image and the cutting track area in the image to be processed may still be relatively sparse, resulting in the inability to filter out the cutting track area in the image to be processed by using the cutting track and unit pattern filtering method; when the first size threshold is too small, even if the size of the unit image is larger than the first size threshold, the distribution of the unit image and the cutting track area in the image to be processed may still be relatively dense, resulting in an increase in the workload of separating the unit image from the image to be processed by using the cutting track avoidance method and reducing the detection efficiency. Therefore, in this embodiment, the range of the first size threshold is 20 to 60 pixel points.

[0073] Please continue to refer to Figure 1, perform step S104. If the size of the unit image is smaller than a preset first size threshold, perform cutting track and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; perform a first recognition process on the first image to be detected, and obtain the pixel points that meet the first threshold condition in the first image to be detected as defective pixel points.

[0074] The size of the unit image being smaller than the preset first size threshold indicates that the unit images and the cutting track regions in the image to be processed are densely distributed. At this time, cutting track and unit pattern filtering processing can be performed on the image to be processed. Specifically, Fourier low-pass filtering processing can be performed to remove the cutting track regions in the image to be processed, thereby avoiding interference to the detection results caused by the cutting track regions.

[0075] Here, the size of the unit image being smaller than the first size threshold specifically means that at least one of the length and width of the unit image is smaller than the first size threshold.

[0076] In this embodiment, by performing frequency-domain low-pass filtering processing on the image to be processed, cutting track and unit pattern filtering processing is performed on the image to be processed to obtain the first image to be detected.

[0077] When the size of the unit image is smaller than the first size threshold, in the image to be processed, the frequency of the defective pixel points is relatively low, while the frequencies of the pixel points in the cutting track regions and the unit patterns in the unit images are relatively high. Therefore, by performing frequency-domain low-pass filtering processing on the image to be processed, the cutting track regions and the unit patterns with relatively high frequencies in the image to be processed are removed.

[0078] Specifically, the steps of performing frequency-domain low-pass filtering processing on the image to be processed include: performing Fourier transform on the image to be processed to obtain a corresponding frequency spectrum matrix; performing position exchange on the diagonal region in the frequency spectrum matrix to obtain a corresponding transferred frequency spectrum matrix; performing low-pass filtering on the transferred frequency spectrum matrix with a preset filtering radius to obtain a corresponding frequency-domain filtering matrix; performing position exchange on the diagonal region of the frequency-domain filtering matrix to obtain a corresponding transferred frequency-domain filtering matrix; performing inverse Fourier transform on the transferred frequency-domain filtering matrix to obtain the first image to be detected.

[0079] Performing Fourier transform on the image to be processed, that is, performing frequency-domain conversion on the image to be processed, to obtain the change of the gray levels of the pixel points in the image to be processed in space. Correspondingly, the frequency spectrum matrix includes the frequency spectrum information of the pixel points in the image to be processed.

[0080] Interchange the positions of the diagonal regions in the spectral matrix, specifically, divide the spectral matrix into four regions centered on the symmetry center on average, and interchange the positions of the diagonal regions in the four regions. For details, please refer to Figure 4 。

[0081] Please refer to Figure 4 , the spectral matrix 40 includes a plurality of elements (not shown), and the plurality of elements are arranged in a matrix array. Divide the elements in the spectral matrix 40 into four regions centered on the symmetry center on average to obtain a first region 401, a second region 402, a third region 403, and a fourth region 404. Among them, the first region 401 and the third region 403 are diagonal regions to each other, and the second region 402 and the fourth region 404 are diagonal regions to each other. Interchange the positions of the diagonal regions in the spectral matrix 40, that is, interchange the positions of the first region 401 and the third region 403, and interchange the positions of the second region 402 and the fourth region 404.

[0082] By interchanging the positions of the diagonal regions in the spectral matrix, the positions of the elements in the top corner region and the center region of the spectral matrix are interchanged.

[0083] In the spectral matrix, the element with zero frequency is located in the top corner region of the spectral matrix, while the element with high frequency is located in the center region of the spectral matrix. Therefore, by interchanging the positions of the diagonal regions of the spectral matrix, the element with zero frequency is transferred to the center region of the matrix, and the element with high frequency is transferred to the top corner region of the matrix, so as to facilitate subsequent low-pass filtering centered on the symmetry center of the matrix.

[0084] Performing low-pass filtering on the transfer spectral transfer matrix with a preset filtering radius means retaining the spectrum within the filtering radius in the transfer spectral transfer matrix and clearing the spectrum outside the filtering radius, so that the low-frequency elements in the transfer spectral transfer matrix are retained and the high-frequency elements are removed. The high-frequency elements correspondingly correspond to the cutting track region and the unit pattern in the image to be processed. Performing low-pass filtering on the transfer spectral transfer matrix with a preset filtering radius can filter out the spectrum of the pixel points in the cutting track region and the unit pattern in the transfer spectral transfer matrix to obtain the frequency-domain filtering matrix.

[0085] It should be noted that the filtering radius should not be too large or too small. If the filtering radius is too large, the spectra of the scribe lines region and the cell patterns cannot be completely filtered out; if the filtering radius is too small, the finally obtained first image to be detected may be relatively blurred, resulting in the inability to accurately identify defective pixel points from the first image to be detected subsequently. Therefore, in this embodiment, the filtering radius is 40 Hz to 70 Hz.

[0086] Swap the positions of the diagonal regions of the frequency-domain filtering matrix, so as to swap the central region and the corner regions of the frequency-domain filtering matrix, so that the elements that were swapped in position in the transfer spectral matrix are restored to the same positions as the elements in the spectral matrix.

[0087] Obtain the transferred frequency-domain filtering matrix. At this point, the low-pass filtering process of the image to be processed has been completed in the frequency domain. By performing an inverse Fourier transform on the transferred frequency-domain filtering matrix, the transferred frequency-domain filtering matrix obtained after the frequency-domain low-pass filtering process is converted into the time domain to obtain the first image to be detected.

[0088] By performing frequency-domain low-pass filtering on the image to be processed, the scribe lines region and the cell patterns in the image to be processed are filtered out, and the first image to be detected obtained is a smooth image with relatively gentle changes in gray values. For details, please refer to Figure 5 .

[0089] Figure 5 Fig. shows a schematic diagram of a first image to be detected obtained by performing frequency-domain low-pass filtering on the image to be processed in an embodiment of the present invention. As Figure 5 shown, in the first image to be detected 500, the changes in pixel values between pixel points are relatively gentle.

[0090] Perform scribe line and cell pattern removal processing on the image to be processed, so that the scribe lines region and the cell patterns in the image to be processed are removed. The effective pixel points in the first image to be detected are only the pixel points in the cell image other than the cell patterns, and because the frequency of defective pixel points is usually less than the filtering radius, the defective pixel points are retained. Therefore, subsequently, by performing a first recognition process on the first image to be detected, the defective pixel points existing in the cell image are obtained, thereby realizing defect detection of the image to be processed.

[0091] The steps of performing the first recognition process on the first image to be detected include: comparing the pixel points in the first image to be detected with the corresponding reference pixel points to obtain a first difference value between the intensity representation values of the pixel points in the first image to be detected and the corresponding reference pixel points; the intensity representation value is related to the sharpness of the first image to be detected; comparing the first difference value with a first threshold value to obtain a comparison result between the first difference value and the first threshold value; and obtaining the defective pixel points based on the comparison result between the first difference value and the first threshold value.

[0092] The reference pixel points serve as a comparison benchmark when performing the first recognition process on the pixel points in the first image to be detected. By comparing the pixel points in the first image to be detected with the corresponding reference pixel points, it is determined whether there are defective pixel points in the first image to be detected.

[0093] Specifically, a first difference value between the intensity representation values of the pixel points in the first image to be detected and the corresponding reference pixel points is obtained to determine whether the pixel points in the first image to be detected are defective pixel points.

[0094] In this embodiment, the currently to-be-detected pixel point in the first image to be detected is used as the to-be-inspected pixel point, and the pixel points in the first image to be detected that are at a preset first distance from the to-be-inspected pixel point are used as the reference pixel points. In other words, with the preset first distance as the radius, the reference pixel points corresponding to the to-be-inspected pixel point are determined.

[0095] In this embodiment, there are multiple reference pixel points corresponding to the to-be-inspected pixel point.

[0096] Figure 6 FIG. shows a schematic diagram of a positional relationship between the to-be-inspected pixel point and the corresponding reference pixel points in an embodiment of the present invention. As Figure 6 shown, as an example, in the first image to be detected 500, there are 8 reference pixel points 525 corresponding to the to-be-inspected pixel point 515. Specifically, the distances between the reference pixel points 525 and the to-be-inspected pixel point 515 are all the first distance S1, and the angles between the connecting lines of the reference pixel points 525 and the to-be-inspected pixel point 515 and the arrangement direction of the pixel points in the first image to be detected 500 are 0°, 45°, 90°, 135°, 180°, 225°, 270°, and 315° respectively. Among them, it is the first arrangement direction (X direction) of the pixel points in the first image to be detected or the second arrangement direction (Y direction) perpendicular to the first arrangement direction.

[0097] The reference pixel point 525 and the pixel point to be inspected 515 are located in the same image to be detected, that is, the reference pixel point 525 and the pixel point to be inspected 515 come from the same object to be measured, thereby avoiding adverse effects on the accuracy of the detection result due to large differences in the average intensity characterization values between different objects to be measured, and correspondingly facilitating the improvement of the accuracy of the detection result.

[0098] The first preset distance S1 can be set according to actual needs. In this embodiment, the first preset distance S1 is 8 to 15 pixel points.

[0099] In other embodiments, the reference pixel point can also come from a template image different from the first image to be detected where the pixel point to be inspected is located.

[0100] Specifically, the template image is an image obtained by photographing a standard object consistent with the object to be measured. It can be understood that the template image and the first image to be detected are not on the same image. Among them, the template image can include a CAD drawing or a defect-free measurement image.

[0101] Correspondingly, perform a first matching process on the first image to be detected and the template image, so that the pixel points in the first image to be detected and the template image correspond one by one, and obtain the reference pixel point corresponding to the pixel point to be inspected in the first image to be detected.

[0102] Here, the reference pixel point corresponding to the pixel point to be inspected is the pixel point at the same position as the pixel point to be inspected in the reference image.

[0103] Comparing the pixel point to be inspected with the reference pixel point to obtain the first difference value between the intensity characterization values of the pixel point to be inspected and the reference pixel point means obtaining the absolute value of the difference between the intensity characterization values of the pixel point to be inspected and the reference pixel point.

[0104] In this embodiment, the intensity characterization value is positively correlated with the gray value or the signal-to-noise ratio. Specifically, the intensity characterization value includes the gray value or the light intensity value or the brightness value of the pixel points forming the first image to be detected.

[0105] In this embodiment, the intensity characterization value is the gray value. Correspondingly, obtain the difference value between the gray value of the pixel point to be inspected in the first image to be detected and the corresponding reference pixel point.

[0106] The difference value between the gray value of the pixel point to be inspected in the first image to be detected and the corresponding reference pixel point correspondingly refers to the absolute value of the difference between the gray value of the pixel point to be inspected in the first image to be detected and the corresponding reference pixel point.

[0107] In this embodiment, there are multiple reference pixel points. Based on the comparison result between the first difference value and the first threshold, the steps of obtaining the defective pixel points include: comparing the pixel points to be detected in the first image to be detected with the corresponding multiple reference pixel points respectively, and obtaining the number of times that the same pixel point to be detected in the first image to be detected is identified as a first abnormal pixel point; if the number of times that the same pixel point to be detected in the first image to be detected is identified as a first abnormal pixel point is greater than the preset first number threshold, taking the pixel point to be detected as the defective pixel point.

[0108] If there are multiple pixel points to be detected in the first image to be detected, comparing the pixel points to be detected in the first image to be detected with one reference pixel point, and when the first difference value between the pixel point to be detected and the reference pixel point is greater than the first threshold, identifying the pixel point to be detected in the first image to be detected as a first abnormal pixel point.

[0109] In the case of comparing with one reference pixel point and the pixel point to be detected in the first image to be detected is identified as a first abnormal pixel point, counting the number of times that the pixel point to be detected in the first image to be detected is identified as a first abnormal pixel point as one time, and so on. If there are multiple reference pixel points corresponding to the pixel point to be detected in the first image to be detected, by comparing the pixel point to be detected with the corresponding multiple reference pixel points respectively, the number of times that the pixel point to be detected in the first image to be detected is determined as a first abnormal pixel point can be obtained.

[0110] For example, please continue to refer to Figure 6 , the number of reference pixel points corresponding to the pixel point to be detected 515 is 8. If the pixel point to be detected 515 is compared with 8 reference pixel points 525 respectively, and relative to 3 of the reference pixel points 525, the pixel point to be detected 515 is identified as the first abnormal pixel point 535, then the number of times that the pixel point to be detected 515 is identified as the first abnormal pixel point 535 is 3 times.

[0111] It can be seen from this that in the case where the number of reference pixel points is multiple, the minimum value of the number of times that the same pixel point to be detected in the first image to be detected is identified as a first abnormal pixel point is zero, and the maximum value is the number of reference pixel points corresponding to the pixel point to be detected.

[0112] Obtain the number of times that the same pixel point to be detected in the first image to be detected is identified as a first abnormal pixel point, and determine whether the first abnormal pixel point is a defective pixel point by comparing the obtained number with the preset first number threshold.

[0113] Specifically, if the obtained number is greater than the preset first number threshold, taking the first abnormal pixel point as the defective pixel point.

[0114] Therefore, the minimum value of the first number threshold is one, and the maximum value is the number of reference pixel points corresponding to the pixel point to be detected.

[0115] It should be noted that the first number threshold should not be too small or too large. If the first number threshold is too small, normal pixel points in the first image to be detected are likely to be identified as defective pixel points, resulting in an increase in the false detection rate and an excessive number of detected defective pixel points, thereby increasing the subsequent data processing volume. If the first number threshold is too large, it is likely to cause the defective pixel points in the first image to be detected to not be accurately identified, increasing the probability of missed detection. For this reason, in this embodiment, when the number of reference pixel points is 8, the first number threshold is 2 to 5 times.

[0116] In other embodiments, the first number threshold is related to the number of reference pixel points. Specifically, the first number threshold is half of the number of reference pixel points.

[0117] By setting the number of reference pixel points for comparing pixel points in the first image to be detected as multiple, the pixel points to be detected in the first image to be detected are respectively compared with multiple reference pixel points, so as to improve the accuracy of the detected defective pixel points, and further provide a basis for accurately obtaining the defective points existing in the image to be detected subsequently.

[0118] In other embodiments, the number of reference pixel points can also be one, and correspondingly, the pixel points with the first difference value greater than the first threshold can also be directly used as defective pixel points to improve the detection efficiency.

[0119] It can be understood that in the case of obtaining the reference pixel points corresponding to the pixel points to be detected from the template image by means of matching processing, the number of reference pixel points is the same as the number of the template images.

[0120] In this embodiment, the first threshold is positively correlated with the sharpness of the first image to be detected. The higher the sharpness of the image, the clearer the image, the more obvious the gray level change at the image contour edge, and the stronger the sense of hierarchy. Therefore, in order to be able to screen out defective pixel points, the value of the first threshold is correspondingly larger.

[0121] The sharpness of the image is related to the gray level gradient. Therefore, the first threshold is obtained through the gray level gradient.

[0122] Specifically, the steps of obtaining the first threshold include: obtaining the gray level gradient of each pixel point in the first image to be detected; obtaining the average value of the gray level gradients of the pixel points in the first image to be detected as the initial threshold; and obtaining the first threshold based on the initial threshold.

[0123] Specifically, based on the initial threshold, obtaining the first threshold includes: using the initial threshold as the first threshold, or increasing the initial threshold by a preset offset to obtain the first threshold.

[0124] It can be understood that in the case where the step of obtaining the first threshold further includes using the initial threshold as the first threshold, the corresponding preset offset of the first threshold is zero accordingly.

[0125] In this embodiment, the first threshold is obtained by increasing the initial threshold by a preset offset.

[0126] It should be noted that the preset offset should not be too small or too large. If the preset offset is too small, during the first recognition process, it is easy to cause a high false detection rate, that is, it is easy to classify normal pixel points as defective pixel points, resulting in too many defective pixel points, and further increasing the data processing amount of obtaining defect points based on the defective pixel points; if the preset offset is too large, the probability of missed detection is likely to increase. Therefore, the preset offset is 3 to 5.

[0127] It should also be noted that any pixel point in the first image to be detected has corresponding gray gradients in the X direction and the Y direction. As an example, calculating the gray gradient of the pixel points in the first image to be detected using the following formula includes:

[0128]

[0129] where M(x, y) represents the gray gradient of the pixel point (x, y) in the first image to be detected, gx represents the gradient of the pixel point (x, y) in the first arrangement direction of the pixel points, and gy represents the gradient of the pixel point (x, y) in the second arrangement direction of the pixel points.

[0130] In other embodiments, different first thresholds can also be set for the pixel points based on the gray values and gray gradients of each pixel point in the first image to be detected, which is not limited here.

[0131] Please continue to refer to Figure 1 , perform step S105. If the size of the unit image is greater than or equal to the first size threshold, perform a cutting track avoidance process on the image to be processed to obtain a second image to be detected; perform a second recognition process on the second image to be detected, and obtain the pixel points that meet the second threshold condition in the second image to be detected as defective pixel points.

[0132] If the size of the unit image is greater than or equal to the first size threshold, it indicates that the distribution of the unit image and the scribe lane region in the image to be processed is relatively sparse, and the scribe lane region in the image to be processed cannot be filtered out by using the frequency domain low-pass filtering method. At this time, the scribe lane avoidance process can be performed on the image to be processed, and each unit image can be segmented from the image to be processed as the second image to be detected, and then each of the segmented second images to be detected can be detected subsequently, so as to avoid the interference of the scribe lane region on the subsequent defect identification.

[0133] Here, the size of the unit image being greater than or equal to the first size threshold specifically means that both the length and width of the unit image are greater than or equal to the first size threshold.

[0134] In this embodiment, the steps of performing the scribe lane avoidance process on the image to be processed include: performing binarization processing on the image to be processed to obtain a binary image; performing pixel value inversion processing on the binary image to obtain an inverted binary image; performing a first connected component judgment on the inverted binary image to obtain the corresponding first connected component; obtaining a reference unit image; the reference unit image has a preset reference size, and the reference size is the standard size of the unit image; comparing the obtained first connected component with the reference unit image, and obtaining the first connected component whose size difference from the reference unit image is less than a preset second size threshold as the valid first connected component; and using the region in the image to be processed corresponding to the valid first connected component as the second image to be detected.

[0135] In this embodiment, the image to be processed is a grayscale image. Performing binarization processing on the image to be detected means comparing the grayscale values of the pixel points in the image to be processed with a preset binarization threshold, setting the pixel values of the pixel points with grayscale values less than the binarization threshold to a first value, and setting the pixel values of the pixel points with grayscale values greater than or equal to the binarization threshold to a second value to obtain the binary image.

[0136] The second value is greater than the first value. In this embodiment, the first value is 0 and the second value is 255. In other embodiments, the first value and the second value can also be other values.

[0137] The binarization threshold can be set according to actual needs. In this embodiment, the steps of obtaining the binarization threshold include: obtaining the maximum value of the grayscale values of the pixel points of the unit image in the image to be processed; and obtaining the binarization threshold based on the obtained maximum value of the grayscale values.

[0138] In this embodiment, by obtaining the histogram of the image to be processed, the gray value corresponding to the highest point in the histogram is obtained, and the maximum value of the gray values of the pixel points of the unit image in the image to be processed is obtained.

[0139] In this embodiment, based on the maximum value of the obtained gray values, obtaining the binarization threshold includes: increasing the maximum value of the obtained gray values by a preset value as the binarization threshold.

[0140] The maximum value of the obtained gray values is increased by a preset value so that when performing binarization conversion subsequently, the set binarization threshold can cover all the gray values of the pixel points in the unit image of the image to be processed, thereby ensuring the accuracy of the binarization conversion.

[0141] In this embodiment, the preset value is from 15 to 30.

[0142] In this embodiment, after obtaining the binarized image, the step of performing cutting track avoidance processing on the image to be processed further includes: performing morphological closing operation on the binarized image to eliminate the noise points existing in the cutting track area.

[0143] By performing morphological closing operation on the binarized image, the interference caused by the noise points existing in the cutting track area to the subsequent identification of defective pixel points can be avoided.

[0144] In this embodiment, the pixel value of the pixel points in the binarized image is 0 or 255. Correspondingly, performing pixel value inversion processing on the binarized image means setting the pixel value of the pixel points with pixel value 0 in the binarized image to 255, and setting the pixel value of the pixel points with pixel value 255 in the binarized image to 0 to obtain the inverse binarized image.

[0145] Figure 7 Shows a partial schematic diagram of an inverse binarized image in an embodiment of the present invention. As Figure 7 shown, in the inverse binarized image 700, the pixel value of the pixel points in the white area 701 is 0, and the pixel value of the pixel points in the black area 702 is 255.

[0146] When the inverse binarized image is obtained, by performing the first connected component judgment on the pixel points in the inverse binarized image, the first connected component existing in the inverse binarized image is obtained. Specifically, the connected component composed of the pixel points with pixel value 0 in the inverse binarized image is obtained as the first connected component.

[0147] The reference unit image is used as a comparison standard for identifying the second image to be detected. By comparing the first connected component obtained in the inverse binarized image with the reference unit image, it serves as the second image to be detected.

[0148] Specifically, by comparing the first connected region identified from the anti-binarized image with the reference unit image respectively, a first connected region whose size difference from the reference image is less than a preset second size threshold is obtained as the second image to be detected.

[0149] It should be noted that in the object to be measured, the sizes of multiple unit structures arranged periodically are the same or substantially the same. In other words, the size difference between multiple unit structures arranged periodically in the object to be measured is less than a preset difference threshold. Correspondingly, the difference between each unit image in the anti-binarized image will also be less than the preset difference threshold. Therefore, a reference unit image is set as a reference standard for the unit image to identify a first connected region in the anti-binarized image whose size difference from the reference unit image is less than the second size threshold as the second image to be detected, that is, the unit image.

[0150] In this embodiment, based on the obtained first connected region, the reference unit image is obtained. Specifically, information on the width and length of each first connected region in the anti-binarized image is obtained; the length and width with the largest quantity among the obtained length and width of the first connected region are respectively used as the length and width of the reference unit image to obtain the reference unit image.

[0151] In other embodiments, the median or average value of the sizes of the unit images identified from the image to be processed can also be used as the size of the reference unit image. Those skilled in the art can make a choice according to the actual situation, as long as the size of the set reference unit image can provide a comparison benchmark for accurately identifying the unit images existing in the anti-binarized image subsequently, which is not limited herein.

[0152] The obtained first connected region is compared with the reference unit image, that is, the sizes between the first connected region and the reference unit image are compared to obtain the size difference between the first connected region in the anti-binarized image and the reference unit image.

[0153] In this embodiment, the sizes of the first connected region and the reference unit image respectively include width and length. Correspondingly, comparing the sizes between the first connected region and the reference unit image means comparing the width and length of the first connected region with the width and length of the reference unit image respectively to obtain the length difference value between the length of the first connected region and the reference unit image and the width difference value between the width of the first connected region and the reference unit image.

[0154] Wherein, the length difference value is the absolute value of the difference between the length of the first connected domain and the length of the reference unit image, and the width difference value is the absolute value of the difference between the width of the first connected domain and the width of the reference unit image.

[0155] If the size between the first connected domain and the reference unit image is smaller than a preset second size threshold, then the corresponding first connected domains are respectively used as the second images to be detected, so as to segment each unit image from the anti-binarized image.

[0156] In this embodiment, the first connected domain with a length difference value less than a preset length threshold and a width difference value less than a preset width threshold from the reference unit image is used as a valid first connected domain. In other words, the second size threshold includes the length threshold and the width threshold.

[0157] The second size threshold can be set according to actual needs. It should be noted that the second size threshold should not be too large or too small. If the second size threshold is too large, it may identify the first connected domain corresponding to a non-unit image as the second image to be detected, and correspondingly increase the data volume for subsequent detection; if the second size threshold is too small, it may miss the first connected domain corresponding to the unit image, thus reducing the accuracy of the detection result. In this embodiment, the second size threshold is 15 to 25 pixel points.

[0158] Please continue to refer to Figure 7 , in the anti-binarized image 700, each white area 701 is respectively a second image to be detected 710, that is, the unit image corresponding to the unit structure of the object to be measured, and the black area 702 correspondingly corresponds to the scribe lane area, that is, the image of the scribe lane of the object to be measured.

[0159] In the case of identifying valid second connected domains from the anti-binarized image, the regions in the image to be processed corresponding to the valid second connected domains are respectively used as the second images to be detected.

[0160] Each of the second images to be detected identified from the image to be processed is the unit image corresponding to each unit structure in the object to be measured. Therefore, subsequent second recognition processing is respectively performed on each of the second images to be detected, which can avoid the scribe lane area, thereby eliminating the interference caused by the scribe lane area to the detection result.

[0161] The steps of performing a second recognition process on the second image to be detected include: obtaining a reference image corresponding to the second image to be detected; matching the second image to be detected with the reference image so that the pixel points in the second image to be detected correspond one by one to those in the reference image; comparing the second image to be detected with the reference image to obtain a second difference value between the intensity characterization values of the corresponding pixel points in the second image to be detected and the reference image; the intensity characterization value is related to the sharpness of the second image to be detected; comparing the obtained second difference value with a second threshold value to obtain a comparison result between the second difference value and the second threshold value; and obtaining the defective pixel points based on the comparison result between the second difference value and the second threshold value.

[0162] The reference image serves as a comparison benchmark when performing a second recognition process on the second image to be detected. By comparing the second image to be detected with the reference image, it is thus determined whether there are defective pixel points in the second image to be detected.

[0163] Specifically, by comparing the difference between the intensity characterization values of the corresponding pixel points in the second image to be detected and the reference image, it is determined whether the pixel points of the second image to be detected are defective pixel points.

[0164] Correspondingly, by performing a matching process on the image to be detected and the reference image, the pixel points of the image to be detected are made to correspond one by one to those of the reference image, and the image to be detected is compared with the reference image to obtain a difference value between the intensity characterization values of the corresponding pixel points in the image to be detected and the reference image.

[0165] The image to be detected and the reference image are in the same image to be detected, that is, the image to be detected and the reference image come from the same object to be measured, thereby avoiding adverse effects on the accuracy of the detection result due to large differences in the average intensity characterization values between different objects to be measured, and correspondingly facilitating the improvement of the accuracy of the detection result.

[0166] Specifically, the currently to-be-detected second image to be detected in the image to be processed is used as the image to be detected, and multiple second images to be detected adjacent to the image to be detected are used as the reference image. As Figure 8 shown, as an example, the number of the reference images 125 is 4, and the image to be detected 115 and the corresponding reference images 125 are arranged in a cross shape.

[0167] It can be understood that the number of the reference images can be more or less, such as 1, 2, or 8, etc. If the number of the reference images is 4, the reference images and the image to be inspected can also be arranged in an X shape; if the number of the reference images is 8, the image to be inspected and the corresponding reference images can also be arranged in a 3*3 array.

[0168] In other embodiments, a standard image can also be used as the reference image. Correspondingly, the image to be inspected and the reference image are subjected to a matching process, so that the pixel points in the matching region between the reference image and the image to be inspected correspond one by one to the pixel points in the image to be detected.

[0169] The reference image is an image of a reference object that is the same as the object to be measured. As an example, the reference image is a CAD drawing of the reference object; as another example, the reference image is a defect-free measurement image of the reference object.

[0170] The image to be inspected is a unit image, and the reference image correspondingly is a reference image of a unit structure.

[0171] Taking the defect-free measurement image as the reference image as an example, by selecting a reference object that is the same as the object to be measured, the reference object also has a plurality of unit structures, obtaining an image of the reference object, and selecting a qualified unit image on the image of the reference object as the reference image. For example, first select a qualified wafer, obtain an image of the selected wafer, and select a qualified die image on the obtained image of the wafer.

[0172] Correspondingly, when performing the second recognition process, the image to be inspected and the reference image are subjected to a matching process, and the image to be inspected and the reference image are compared to obtain a second difference value between the intensity characterization values of the corresponding pixel points in the image to be processed and the reference image.

[0173] The corresponding pixel points in the image to be inspected and the reference image refer to the pixel points at the same positions in the image to be inspected and the reference image.

[0174] Comparing the image to be inspected and the reference image to obtain a second difference value between the intensity characterization values of the corresponding pixel points in the image to be inspected and the reference image means obtaining the absolute value of the difference between the intensity characterization values of the corresponding pixel points in the image to be inspected and the reference image.

[0175] In this embodiment, the intensity characterization value is a gray value. Correspondingly, obtaining the difference value between the intensity characterization values of the corresponding pixel points in the image to be inspected and the reference image means obtaining the difference value between the gray values of the corresponding pixel points in the image to be inspected and the reference image.

[0176] Among them, the difference value between the gray values of the corresponding pixel points in the image to be inspected and the reference image correspondingly refers to the absolute value of the difference between the gray values of the corresponding pixel points in the image to be inspected and the reference image.

[0177] In this embodiment, there are multiple reference images. Based on the comparison result between the second difference value and the second threshold, the steps of obtaining the defective pixel points include: comparing the image to be inspected with the multiple reference images respectively, and obtaining the number of times that the same pixel point in the image to be inspected is identified as a second abnormal pixel point; if the obtained number is greater than the preset second number threshold, the second abnormal pixel point is used as the defective pixel point.

[0178] Compare the image to be inspected with one reference image, and when the second difference value between the gray values of the corresponding pixel points in the image to be inspected and the reference image is greater than the second threshold, the pixel point in the image to be inspected is identified as a second abnormal pixel point.

[0179] When the image to be inspected is compared with one reference image and the pixel point in the image to be inspected is identified as a second abnormal pixel point, the number of times that the pixel point is identified as a second abnormal pixel point is counted as one, and so on. If there are multiple reference images corresponding to the image to be inspected, by comparing the image to be inspected with the multiple reference images respectively, the number of times that the same pixel point in the image to be inspected is determined to be a second abnormal pixel point can be obtained.

[0180] Obtain the number of times that the same pixel point in the image to be inspected is identified as a second abnormal pixel point, and compare the obtained number with the preset second number threshold to determine whether the second abnormal pixel point is a defective pixel point. Specifically, if the obtained number is greater than the preset second number threshold, the second abnormal pixel point is used as the defective pixel point. Among them, the minimum value of the second number threshold is one, and the maximum value is the number of reference images corresponding to the image to be inspected.

[0181] The second number threshold can be executed with reference to the foregoing first number threshold, which will not be elaborated here. It can be understood that the second number threshold and the second number threshold may be the same or different, and those skilled in the art can select according to actual needs, which is not limited here.

[0182] By setting the number of reference images for comparing the image to be inspected to be multiple, the image to be inspected is compared with the multiple reference images respectively, so as to improve the accuracy of the detected defective pixel points, and further provide a basis for accurately obtaining the defective points existing in the image to be inspected subsequently.

[0183] In other embodiments, the number of reference images corresponding to the image to be inspected can also be one. Correspondingly, the pixel points with the second difference value greater than the second threshold can also be directly used as defective pixel points to improve the detection efficiency.

[0184] Please continue to refer to Figure 1 , perform step S106, and based on the defective pixel points, obtain the defective points existing in the image to be processed.

[0185] In the foregoing, the corresponding recognition process is performed on the first image to be detected or the second image to be detected in units of pixel points, and the pixel points that meet the corresponding threshold conditions are obtained as defective pixel points. However, the detected defective pixel points may still be noise points, and the size of the defective points is not limited to one pixel point, but also includes the case where multiple defective pixel points are connected.

[0186] The step of obtaining the defective points existing in the image to be processed based on the defective pixel points includes: performing a second connected component judgment on the defective pixel points to obtain the corresponding second connected component; if the corresponding second connected component is obtained, taking the obtained second connected component as the first candidate defective point; comparing the size of the first candidate defective point with a preset third size threshold, and obtaining the first candidate defective point with a size greater than the preset third size threshold as the second candidate defective point; performing clustering processing on the second candidate defective points, taking the second candidate defective points with a distance less than the preset distance threshold as one defective point, and taking the second candidate defective points with a distance greater than the distance threshold from other second candidate defective points as another defective point.

[0187] The size of noise points is usually small. By performing a second connected component judgment, the isolated defective pixel points are screened out as noise points.

[0188] Specifically, if the corresponding second connected component is obtained, the obtained second connected component is taken as the first candidate defective point; if the corresponding second connected component is not obtained, the defective pixel points are screened out as noise points. In other words, when the defective pixel points are isolated single pixel points, the defective pixel points are taken as noise points; otherwise, all the defective pixel points in the connected component are taken as a whole as the first candidate defective point.

[0189] In this embodiment, the second connected component judgment includes a four-connected component judgment or an eight-connected component judgment. Among them, the four-connected component judgment means: judging whether there are 4 adjacent defective pixel points around any defective pixel point; the eight-connected component judgment means: judging whether there are 8 adjacent defective pixel points around any defective pixel point.

[0190] Compare the size of the first candidate defect point with a preset third size threshold, and filter out the first candidate defect points with a size smaller than the preset third size threshold as noise points.

[0191] The first candidate defect point includes at least two adjacent defective pixel points, and the size of the first candidate defect point correspondingly includes at least one of parameters such as length, width, radius, area, etc. Therefore, the third size threshold includes at least one of the thresholds of parameters such as length, width, radius, area, etc.

[0192] Cut the image to be processed according to the size of the unit image therein for cutting channel and unit pattern filtering or cutting channel avoidance processing. Some defects (such as scratches) existing in the image to be processed may span several unit images. Therefore, in order to detect defects spanning several unit images, clustering processing is performed on the detected second candidate defect points.

[0193] In this embodiment, a spatial distance clustering algorithm is used to perform clustering processing on the second candidate defect points. Specifically, the second candidate defect points with a distance less than a preset distance threshold are regarded as one defect point, and the second candidate defect points with a distance greater than or equal to the distance threshold from other second candidate defect points are regarded as one defect point.

[0194] Here, the second candidate defect points with a distance less than the preset distance threshold mean that the distance between any two second candidate defect points is less than the distance threshold.

[0195] The distance threshold can be set according to actual needs. It can be understood that the distance threshold should not be too large or too small. If the distance threshold is too large, second candidate defect points that do not belong to one defect point may be grouped into one defect point; if the distance threshold is too small, second candidate defect points belonging to the same defect point may be split. For this reason, in this embodiment, the distance threshold is 20 to 100 pixel points.

[0196] In other embodiments, the second defect candidate points can also be directly regarded as defect points.

[0197] Correspondingly, an embodiment of the present invention also provides a detection device.

[0198] Figure 9 The structural schematic diagram of a detection device in an embodiment of the present invention is shown. Please refer to Figure 9, the detection device 90 may include: an image acquisition unit 901, adapted to acquire an image to be processed; the image to be processed includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane region; the unit image includes a unit pattern; a size acquisition unit 902, adapted to acquire the size of the unit image in the image to be processed; a size comparison unit 903, adapted to compare the size of the unit image with a preset first size threshold; an identification processing unit 904, adapted to, if the size of the unit image is less than the first size threshold, perform scribe lane and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; perform first identification processing on the first image to be detected to obtain pixel points that meet the first threshold condition in the first image to be detected as defective pixel points; if the size of the unit image is greater than or equal to the first size threshold, perform scribe lane avoidance processing on the image to be processed to obtain a second image to be detected; perform second identification processing on the second image to be detected to obtain pixel points that meet the second threshold condition in the second image to be detected as defective pixel points; a defect acquisition unit 905, adapted to obtain defect points existing in the image to be processed based on the defective pixel points.

[0199] The detection device is used to execute the detection method described in the foregoing embodiment, and other structures may also be used to execute the detection method described in the foregoing embodiment. For a specific description of the detection device in this embodiment, reference may be made to the corresponding description of the detection method in the foregoing embodiment, which will not be elaborated herein.

[0200] An embodiment of the present invention further provides a device, and the device can implement the detection method provided by the embodiment of the present invention by loading the above detection method in the form of a program.

[0201] Reference Figure 10 , which shows a hardware structure diagram of the device provided by an embodiment of the present invention. The device in this embodiment includes: at least one processor 01, at least one communication interface 02, at least one memory 03, and at least one communication bus 04.

[0202] In this embodiment, the number of the processor 01, the communication interface 02, the memory 03, and the communication bus 04 is at least one, and the processor 01, the communication interface 02, and the memory 03 complete communication with each other through the communication bus 04.

[0203] The communication interface 02 may be an interface of a communication module for network communication, for example, an interface of a GSM module.

[0204] The processor 01 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the detection method described in this embodiment.

[0205] The memory 03 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory.

[0206] Among them, the memory 03 stores one or more computer instructions, and the one or more computer instructions are executed by the processor 01 to implement the detection method provided in the foregoing embodiment.

[0207] It should be noted that the above-mentioned implementation terminal device may also include other devices (not shown) that may not be necessary for the disclosed content of the embodiments of the present invention; in view of the fact that these other devices may not be necessary for understanding the disclosed content of the embodiments of the present invention, the embodiments of the present invention will not introduce them one by one.

[0208] The embodiments of the present invention also provide a storage medium, which stores one or more computer instructions, and the one or more computer instructions are used to implement the detection method provided in the foregoing embodiment.

[0209] The above embodiments of the present invention are combinations of elements and features of the present invention. Unless otherwise mentioned, the elements or features can be regarded as optional. Each element or feature can be practiced without being combined with other elements or features. In addition, the embodiments of the present invention can be constructed by combining some elements and / or features. The operation sequence described in the embodiments of the present invention can be rearranged. Some configurations of any embodiment can be included in another embodiment and can be replaced by the corresponding configuration of another embodiment. It is obvious to those skilled in the art that the claims that do not have an explicit citation relationship with each other in the appended claims can be combined into the embodiments of the present invention, or can be included as new claims in the amendments after the submission of this application.

[0210] The embodiments of the present invention can be implemented by various means such as hardware, firmware, software, or a combination thereof. In the hardware configuration mode, the method according to the exemplary embodiments of the present invention can be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, etc.

[0211] In a firmware or software configuration mode, embodiments of the present invention may be implemented in the form of modules, procedures, functions, etc. Software code can be stored in a memory unit and executed by a processor. The memory unit is located inside or outside the processor and can send data to and receive data from the processor via various known means.

[0212] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0213] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be subject to the scope defined by the claims.

Claims

1. A detection method, characterized in that, Including: Obtain an image to be processed; the image to be processed includes a plurality of unit images, and the plurality of unit images are separated from each other by a scribe lane region; the unit image includes a unit pattern. Obtain the size of the unit image in the image to be processed. Compare the size of the unit image in the image to be processed with a preset first size threshold. If the size of the unit image is smaller than the first size threshold, perform scribe lane and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; perform first recognition processing on the first image to be detected, and obtain the pixel points that meet the first threshold condition in the first image to be detected as defective pixel points. The performing scribe lane and unit pattern filtering processing on the image to be processed includes: Perform Fourier low-pass filtering processing on the image to be processed to remove the scribe lane region in the image to be processed to obtain the first image to be detected. The performing first recognition processing on the first image to be detected and obtaining the pixel points that meet the first threshold condition in the first image to be detected as defective pixel points includes: Compare the pixel points in the first image to be detected with the corresponding reference pixel points, and obtain a first difference value between the intensity characterization values of the pixel points in the first image to be detected and the corresponding reference pixel points; the intensity characterization value is related to the sharpness of the first image to be detected. Compare the first difference value with a first threshold to obtain a comparison result between the first difference value and the first threshold; the first threshold is related to the sharpness of the first image to be detected. Obtain the defective pixel points based on the comparison result between the first difference value and the first threshold. If the size of the unit image is greater than or equal to the first size threshold, perform scribe lane avoidance processing on the image to be processed to obtain a second image to be detected; perform second recognition processing on the second image to be detected, and obtain the pixel points that meet the second threshold condition in the second image to be detected as defective pixel points. The performing scribe lane avoidance processing on the image to be processed includes: Filter out the scribe lane region in the image to be processed by using frequency domain low-pass filtering processing to obtain the second image to be detected. The performing second recognition processing on the second image to be detected includes: Obtain a reference image corresponding to the second image to be detected; the reference image is a standard image of a unit structure. Perform matching processing on the second image to be detected and the reference image so that the pixel points in the second image to be detected and the reference image correspond one by one. Compare the second image to be detected with the reference image, and obtain a second difference value between the intensity characterization values of the corresponding pixel points in the second image to be detected and the reference image; the intensity characterization value is related to the sharpness of the second image to be detected. Compare the obtained second difference value with a second threshold to obtain a comparison result between the second difference value and the second threshold; the second threshold is related to the sharpness of the second image to be detected. Based on the comparison result between the second difference value and the second threshold, obtain the defective pixel points; Based on the defective pixel points, obtain the defect points existing in the image to be processed.

2. The detection method according to claim 1, characterized in that, The first size threshold is 20 to 60 pixel points.

3. The detection method according to claim 1, characterized in that, The performing of the scribe lane and cell pattern filtering process on the image to be processed includes: Perform Fourier transform on the image to be processed to obtain the corresponding frequency spectrum matrix; Exchange the positions of the diagonal regions in the frequency spectrum matrix to obtain the corresponding transferred frequency spectrum matrix; perform low-pass filtering on the transferred frequency spectrum matrix with a preset filtering radius to obtain the corresponding frequency domain filtering matrix; Exchange the positions of the diagonal regions in the frequency domain filtering matrix to obtain the corresponding transferred frequency domain filtering matrix; Perform inverse Fourier transform on the transferred frequency domain filtering matrix to obtain the first image to be detected.

4. The detection method according to claim 1, characterized in that, There is one reference pixel point; the obtaining of the defective pixel points based on the comparison result between the first difference value and the first threshold includes: If the first difference value is greater than the first threshold, use the corresponding pixel point in the first image to be detected as the defective pixel point.

5. The detection method according to claim 1, characterized in that, There are multiple reference pixel points; the obtaining of the defective pixel points based on the comparison result between the first difference value and the first threshold includes: If the pixel points in the first image to be detected are respectively compared with the corresponding multiple reference pixel points, obtain the number of times the same pixel point in the first image to be detected is identified as the first abnormal pixel point; wherein, if the pixel point in the first image to be detected is compared with one reference pixel point and the first difference value is greater than the first threshold, the number of times the corresponding pixel point in the first image to be detected is identified as the abnormal pixel point is counted as one time; If the number of times the same pixel point in the first image to be detected is identified as the first abnormal pixel point is greater than the first number threshold, use the corresponding pixel point in the first image to be detected as the defective pixel point.

6. The detection method according to claim 1, characterized in that, The performing of the scribe lane avoidance process on the image to be processed includes: Perform binarization processing on the image to be detected to obtain a binarized image; Perform pixel value inversion processing on the binarized image to obtain an inverse binarized image; Perform the first connected component judgment on the inverse binarized image to obtain the corresponding first connected component; Obtain a reference cell image; the reference cell image has a preset reference size, and the reference size is the standard size of the cell image; Compare the obtained first connected component with the reference cell image, and obtain the first connected component whose size difference from the reference cell image is less than the preset second size threshold as the valid first connected component; Obtain the region in the image to be processed corresponding to the valid first connected component as the second image to be detected.

7. The detection method according to claim 6, wherein, The obtaining of the reference cell image includes: based on the obtained first connected component, obtain the reference cell image.

8. The detection method according to claim 7, wherein, The sizes of the first connected component and the reference cell image respectively include width and length; The obtaining of the reference cell image based on the obtained first connected component includes: Take the length and width with the largest quantity among the length and width of the obtained first connected component as the length and width of the reference unit image respectively, and obtain the reference unit image.

9. The detection method according to claim 1, wherein, There is one reference image; obtaining the defective pixel points based on the comparison result between the second difference value and the second threshold includes: If the second difference value is greater than the second threshold, take the corresponding pixel point in the second image to be detected as the defective pixel point.

10. The detection method according to claim 1, wherein, There are multiple reference images; obtaining the defective pixel points based on the comparison result between the second difference value and the second threshold includes: If the second difference value is greater than the second threshold, take the pixel point in the second image to be detected as the second abnormal pixel point; compare the second image to be detected with the multiple reference images respectively, and obtain the number of times the same pixel point in the image to be detected is identified as the second abnormal pixel point; If the number of times the same pixel point in the second image to be detected is identified as the second abnormal pixel point is greater than the second number threshold, take the corresponding pixel point in the second image to be detected as the defective pixel point.

11. The detection method according to claim 1, wherein, Obtaining the defective points existing in the image to be processed based on the defective pixel points includes: Perform a second connected component judgment on the defective pixel points; If a corresponding second connected component is obtained, take the second connected component as the first candidate defective point; If no corresponding connected component is obtained, take the defective pixel points as the first candidate defective points; Obtain the first candidate defective points with a size greater than a preset third size threshold as the second candidate defective points; Perform clustering processing on the second candidate defective points, take the second candidate defective points with a distance less than the preset distance threshold as one defective point, and take the second candidate defective points with a distance greater than the distance threshold from other second candidate defective points as another defective point.

12. A detection device, wherein, Including: An image acquisition unit adapted to acquire an image to be processed; the image to be processed includes multiple unit images, and the multiple unit images are separated from each other by a scribe lane region; the unit image includes a unit pattern; A size acquisition unit adapted to acquire the size of the unit image in the image to be processed; A size comparison unit compares the size of the unit image in the image to be processed with a preset first size threshold; An identification processing unit is adapted to, if the size of the unit image is less than the first size threshold, perform scribe lane and unit pattern filtering processing on the image to be processed to obtain a first image to be detected; perform a first identification processing on the first image to be detected, and take the pixel points satisfying the first threshold condition in the first image to be detected as defective pixel points; the performing scribe lane and unit pattern filtering processing on the image to be processed includes: Perform Fourier low-pass filtering processing on the image to be processed to remove the scribe lane region in the image to be processed to obtain the first image to be detected; The performing a first identification processing on the first image to be detected and taking the pixel points satisfying the first threshold condition in the first image to be detected as defective pixel points includes: Compare the pixel points in the first image to be detected with the corresponding reference pixel points to obtain a first difference value between the intensity characterization values of the pixel points in the first image to be detected and the corresponding reference pixel points; the intensity characterization value is related to the sharpness of the first image to be detected; Compare the first difference value with a first threshold to obtain a comparison result between the first difference value and the first threshold; the first threshold is related to the sharpness of the first image to be detected; Based on the comparison result between the first difference value and the first threshold, obtain the defective pixel points; if the size of the unit image is greater than or equal to the first size threshold, perform a cutting track avoidance process on the image to be processed to obtain a second image to be detected; perform a second recognition process on the second image to be detected, and obtain the pixel points that meet the second threshold condition in the second image to be detected as the defective pixel points; the performing a cutting track avoidance process on the image to be processed includes: Using a frequency domain low-pass filtering method to filter out the cutting track area in the image to be processed to obtain the second image to be detected; The performing a second recognition process on the second image to be detected includes: Obtain a reference image corresponding to the second image to be detected; the reference image is a standard image of the unit structure; Perform a matching process on the second image to be detected and the reference image so that the pixel points in the second image to be detected and the reference image correspond one by one; Compare the second image to be detected with the reference image to obtain a second difference value between the intensity characterization values of the corresponding pixel points in the second image to be detected and the reference image; the intensity characterization value is related to the sharpness of the second image to be detected; Compare the obtained second difference value with a second threshold to obtain a comparison result between the second difference value and the second threshold; the second threshold is related to the sharpness of the second image to be detected; Based on the comparison result between the second difference value and the second threshold, obtain the defective pixel points; A defect acquisition unit, adapted to obtain the defect points existing in the image to be processed based on the defective pixel points.

13. A computer device, characterized in that, Comprising at least one memory and at least one processor, the memory stores one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the detection method according to any one of claims 1 to 11.

14. A storage medium, characterized in that, The storage medium stores one or more computer instructions, and the one or more computer instructions are used to implement the detection method according to any one of claims 1 to 11.

Images