Angle measurement method and semiconductor measurement device

Abstract

The invention discloses an angle measurement method and a semiconductor measurement device, and the method comprises the steps: firstly obtaining image data of targets in an overall distribution region in a substrate and preset points with representativeness and consistency on each target, and then determining a slice image with a fixed size in the image data with each preset point as a center, the method comprises the following steps of: obtaining a plurality of slice images, calculating deviation values of adjacent slice images, then adjusting coordinates of preset points based on the deviation values to obtain target points, enabling the target points to accurately reflect the consistency of relative positions on a target, finally carrying out straight line fitting on the target points, and determining an included angle between a fitting straight line and an ideal arrangement direction. According to the method, high-precision angle calculation is achieved, the angle measurement method has better stability for substrates of different sizes and the condition that the overall distribution areas of targets in the substrates are different, and a solid technical support is provided for the precision of the manufacturing process and the stability of the product quality.

Description

Technical Field

[0001] The present application relates to the field of precision detection technology, and in particular to an angle measurement method and a semiconductor measurement device. Background Art

[0002] In fields such as industrial manufacturing and precision inspection, calculating the angle between a target and a specific direction (such as a vertical line) is crucial. As generative manufacturing evolves towards high precision and automation, higher requirements are placed on the accuracy, real-time nature, and adaptability of angle calculations.

[0003] In related technologies, image matching and coordinate fitting technology is one of the important means to realize angle calculation. By processing and analyzing the target in the image and obtaining its geometric features, the angle information can be calculated.

[0004] However, the existing angle calculation methods have poor measurement accuracy and stability. Summary of the Invention

[0005] In view of this, the present application provides an angle measurement method and a semiconductor measurement device, which significantly improve the accuracy and stability of angle measurement.

[0006] To solve the above problems, the technical solutions provided by this application are as follows:

[0007] In one aspect, the present application provides an angle measurement method, comprising:

[0008] Acquire image data of an overall distribution area of ​​N targets in a substrate and N preset points, wherein the N preset points are respectively located on the N targets and aligned along an ideal arrangement direction of the N targets, and the spacing between adjacent preset points in the N preset points is determined according to the spacing between adjacent targets;

[0009] Determining N slice images of fixed sizes in the image data with the N preset points as centers, respectively, the N slice images being used to identify local information of the N targets;

[0010] Calculate the offset value of the mth slice image relative to the m-1th slice image using a preset image matching algorithm to obtain N-1 offset values;

[0011] Accumulate the N-1 offset values ​​to the N preset points to obtain N target points;

[0012] A straight line fitting is performed based on the N target points, and an angle between the fitting straight line and the ideal arrangement direction is determined.

[0013] In a possible implementation, calculating the offset value of the m-th slice image relative to the m-1-th slice image to obtain N-1 offset values ​​includes:

[0014] Performing discrete Fourier transform on the N slice images respectively to obtain corresponding frequency domain information;

[0015] A phase correlation matching algorithm is executed based on the frequency domain information to determine the offset value of the m-th slice image relative to the m-1-th slice image, thereby obtaining N-1 offset values.

[0016] In a possible implementation, the image matching algorithm includes one of an image matching algorithm based on a sum of squared differences, an image matching algorithm based on normalized correlation, an image matching algorithm based on deep learning, or a phase correlation matching algorithm.

[0017] In a possible implementation, the image data includes multiple image frames scanned according to the ideal arrangement direction, and the N slice images of fixed sizes determined in the image data with the N preset points as centers respectively include:

[0018] For the i-th preset point, determining the j-th image frame corresponding to the i-th preset point;

[0019] Calculate a rectangular area with a fixed size, taking the i-th preset point as the center;

[0020] If the rectangular area does not exceed the boundary of the j-th image frame, extracting image data corresponding to the rectangular area in the j-th image frame to obtain a slice image corresponding to the i-th preset point;

[0021] If the rectangular area exceeds the boundary of the image frame, image data corresponding to the rectangular area is extracted from the j-th image frame and adjacent image frames to obtain a slice image corresponding to the i-th preset point.

[0022] In a possible implementation, performing straight line fitting based on the N target points and determining an angle between the fitting straight line and the ideal arrangement direction includes:

[0023] Fitting the N target points using the least squares method to obtain a fitting straight line;

[0024] The angle between the fitting straight line and the ideal arrangement direction is determined.

[0025] In a possible implementation, the target is a chip unit, and the substrate is a wafer.

[0026] In another aspect, the present application provides an angle measuring device, comprising a carrying unit, an illumination unit, an imaging unit, and a processing unit:

[0027] The carrying unit is used to carry the object to be tested;

[0028] The lighting unit is used to illuminate the object to be tested;

[0029] The imaging unit is used to obtain image data of the entire distribution area of ​​N targets on the object to be measured in the substrate and N preset points, wherein the N preset points are respectively located on the N targets and aligned along the ideal arrangement direction of the N targets, and the spacing between adjacent preset points in the N preset points is determined according to the spacing between adjacent targets;

[0030] The processing unit includes a slice image determination unit, an offset value calculation unit, an accumulation unit, and an angle calculation unit:

[0031] The slice image determination unit is configured to determine N slice images of fixed sizes in the image data, with the N preset points as centers, respectively, and the N slice images are used to identify local information of the N targets;

[0032] The offset value calculation unit is used to calculate the offset value of the m-th slice image relative to the m-1-th slice image by using the Yisuhe image matching algorithm to obtain N-1 offset values;

[0033] The accumulating unit is configured to accumulate the N-1 offset values ​​to the N preset points to obtain N target points;

[0034] The angle calculation unit is used to perform straight line fitting based on the N target points and determine the angle between the fitted straight line and the ideal arrangement direction.

[0035] In a possible implementation, the image matching algorithm includes one of an image matching algorithm based on a sum of squared differences, an image matching algorithm based on normalized correlation, an image matching algorithm based on deep learning, or a phase correlation matching algorithm.

[0036] In a possible implementation, the offset calculation unit is specifically configured to:

[0037] Performing Fourier transform on the N slice images respectively to obtain corresponding frequency domain information;

[0038] Matching is performed based on the frequency domain information to determine the offset value of the m-th slice image relative to the m-1-th slice image, thereby obtaining N-1 offset values.

[0039] In a possible implementation, the image data includes a plurality of image frames scanned according to the ideal arrangement direction, and the image determination unit is specifically configured to:

[0040] For the i-th preset point, determining the j-th image frame corresponding to the i-th preset point;

[0041] Calculate a rectangular area with a fixed size, taking the i-th preset point as the center;

[0042] If the rectangular area does not exceed the boundary of the j-th image frame, extracting image data corresponding to the rectangular area in the j-th image frame to obtain a slice image corresponding to the i-th preset point;

[0043] If the rectangular area exceeds the boundary of the image frame, image data corresponding to the rectangular area is extracted from the j-th image frame and adjacent image frames to obtain a slice image corresponding to the i-th preset point.

[0044] In a possible implementation, the angle calculation unit is specifically configured to:

[0045] Fitting the N target points using the least squares method to obtain a fitting straight line;

[0046] The angle between the fitting straight line and the ideal arrangement direction is determined.

[0047] In a possible implementation, the target is a chip unit, and the substrate is a wafer.

[0048] On the other hand, the present application provides a computer-readable storage medium, which is used to store a computer program. When the computer program is executed by a computer device, it implements any of the methods described above.

[0049] The beneficial effects of this application are:

[0050] This technical solution is used to precisely measure the angle between the overall distribution position of multiple targets in a substrate and the ideal arrangement direction. The targets in the substrate are arranged in sequence at a fixed spacing and have the same shape and structure. The computer device first obtains the image data of the overall distribution area of ​​N targets in the substrate and N preset points, wherein the N preset points are respectively located on the N targets and aligned along the ideal arrangement direction of the targets, and the spacing between adjacent preset points is determined according to the spacing between adjacent targets, so that these preset points are representative and consistent on each target. Then, N slice images of fixed size are determined in the image data with the N preset points as the center, which are used to identify the local information of the N targets. Since the targets have the same structure, the N slice images can be image registered, and the computer can calculate the image data. The offset value of the mth slice image relative to the m-1th slice image is calculated to obtain N-1 offset values, so that adjacent slice images are aligned through local information, which can effectively avoid the influence of interference information of similar structures inside the target. Then, the N-1 offset values ​​are correspondingly accumulated to N preset points to obtain N target points, so that these target points can accurately reflect the consistency of relative positions on the target. Finally, a straight line is fitted to the N target points, and the angle between the fitted straight line and the ideal arrangement direction is determined, thereby achieving high-precision angle measurement. In addition, this angle measurement method has better stability for substrates of different sizes and different overall distribution areas of the target in the substrate, providing solid technical support for the accuracy of the manufacturing process and the stability of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 A flowchart of an angle measurement method provided in an embodiment of the present application;

[0053] Figure 2 A schematic structural diagram of a wafer provided in an embodiment of the present application;

[0054] Figure 3 A data flow diagram of an angle measurement method provided in an embodiment of the present application;

[0055] Figure 4 A schematic diagram of an angle measurement device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to help those skilled in the art better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0057] As mentioned in the background, image matching and coordinate fitting are important methods for angle calculation. By processing and analyzing the target in the image and obtaining its geometric features, angle information can be calculated. However, existing angle measurement methods have the following problems:

[0058] (1) Traditional image matching algorithms, such as template matching, are easily affected by image illumination changes and noise interference, resulting in large matching errors, which in turn affects the accuracy of angle calculation. The errors are even more significant in the measurement of tiny structures or complex background images.

[0059] (2) For target structures of different sizes and shapes, existing methods often need to be customized for specific scenarios and lack versatility. When the position and size of the target structure in the image change, it is difficult to ensure the stability and accuracy of the measurement.

[0060] (3) Some angle measurement methods rely on manual intervention, resulting in a low degree of automation. For example, manual selection of measurement areas and adjustment of parameters cannot meet the needs of large-scale automated production, resulting in low measurement efficiency and high labor costs.

[0061] The present application provides an angle measurement method and a semiconductor measurement device for calculating the angle between the overall distribution position of multiple targets in a substrate and the ideal arrangement direction. The targets in the substrate are arranged in a fixed spacing order and have the same shape and structure. The computer device first obtains image data of the overall distribution area of ​​N targets in the substrate and N preset points, wherein the N preset points are respectively located on the N targets and aligned along the ideal arrangement direction of the targets, and the spacing between adjacent preset points is determined according to the spacing between adjacent targets, so that these preset points are representative and consistent on each target. Then, N slice images of fixed size are determined in the image data with the N preset points as the center, which are used to identify the local information of the N targets. Since the targets have the same structure, the N slice images can be Image registration is performed, and the offset value of the mth slice image relative to the m-1th slice image is calculated to obtain N-1 offset values, so that adjacent slice images are registered through local information, which can effectively avoid the influence of interference information of similar structures inside the target. Then, the N-1 offset values ​​are correspondingly accumulated to N preset points to obtain N target points, so that these target points can accurately reflect the consistency of the relative position on the target. Finally, a straight line is fitted to the N target points, and the angle between the fitted straight line and the ideal arrangement direction is determined, achieving high-precision angle measurement. This angle measurement method has better stability for substrates of different sizes and different overall distribution areas of the target in the substrate, providing solid technical support for the accuracy of the manufacturing process and the stability of product quality.

[0062] The solutions provided in the embodiments of the present application relate to the field of precision detection technology, and are specifically described through the following embodiments.

[0063] See also Figure 1 As shown, it is a flowchart of an angle measurement method provided in an embodiment of the present application. In this embodiment, it can be executed by a computer device as an example.

[0064] S201: Acquire image data of an overall distribution area of ​​N targets in a substrate and N preset points.

[0065] The N preset points are respectively located on the N targets and aligned along an ideal arrangement direction of the N targets. The spacing between adjacent preset points among the N preset points is determined according to the spacing between adjacent targets.

[0066] The ideal alignment is the reference direction for precise alignment of targets on a substrate according to design rules. In many scenarios, such as industrial manufacturing and semiconductor testing, there is a certain deviation between the actual alignment and the ideal alignment. This deviation angle needs to be calculated to evaluate product performance and quality, ensuring overall quality and functional consistency.

[0067] Image data can be obtained by scanning the entire distribution area of ​​N targets on the substrate. To ensure data consistency, N preset points are identified in the image data with coordinate information to obtain coordinate information for the N preset points. A computer device can receive the image data with consistent coordinate information and the N preset points for subsequent calculations.

[0068] In one possible implementation, the target is a chip unit and the substrate is a wafer. Of course, the substrate can also be a semiconductor product such as a PCB board or a solar panel, and the target is the functional components arranged and distributed on these semiconductor products.

[0069] refer to Figure 2 The figure shows a schematic diagram of the structure of a wafer proposed in an embodiment of the present application. In this schematic diagram, five chip units are arranged sequentially from left to right, and the shape and structure of each chip unit are consistent. To ensure the precision of the wafer, it is necessary to calculate the angle between the overall distribution position of the chip units and their ideal arrangement direction, that is, to calculate the angle a, so as to evaluate and optimize the alignment accuracy during the wafer manufacturing process, thereby improving the overall quality and performance of the wafer.

[0070] by Figure 2 The wafer structure shown in Figure 1 illustrates this: the overall distribution area of ​​N chip units within the wafer is a strip at the center of the wafer. Scanning technology can capture image data of this strip. The ideal arrangement of the chip units is from left to right. Assuming the first preset point is located at the center of the first chip unit, a rectangular coordinate system is established with this point as the origin and the ideal arrangement direction along the x-axis. The spacing between adjacent chip units is 1. The coordinates of the other N preset points are (1, 0), (2, 0)…(N-1, 0).

[0071] S202: Determine N slice images of fixed sizes in the image data with N preset points as centers respectively.

[0072] The area of ​​the slice image is smaller than the area of ​​a single target and is used to identify the local information of the target.

[0073] Existing techniques typically use a fixed-grid slicing strategy, which divides the original image into a fixed-size grid without relying on a preset point set. This method is simple and computationally inefficient; however, it fails to optimize the target structure, potentially resulting in excessive amounts of useless information in the sliced ​​image, impacting processing efficiency and accuracy. This strategy is suitable for scenarios where the target structure is complex and variable, making it difficult to accurately locate using a preset point set, and where high slicing accuracy is required.

[0074] The strategy of slicing according to preset points in the embodiment of the present application is applicable to the situation where the structure and position range of the target are known, and the slice image can be accurately positioned through the preset point set.

[0075] After obtaining the image data and the coordinate information of the preset points, each of the N preset points is taken as the center, and region calculation is performed according to a fixed size, and the image data is sliced ​​to obtain slice images corresponding to the N preset points.

[0076] Image data may be composed of multiple images, and the position of the target in each image is not consistent. Slicing operations need to be performed according to the actual situation to obtain a slice image representing the local information of the target on each target.

[0077] In a possible implementation, the image data includes multiple image frames scanned in an ideal arrangement direction, and step S202 includes:

[0078] S2021: For the i-th preset point, determine the j-th image frame corresponding to the i-th preset point.

[0079] S2022: Calculate a rectangular area with a fixed size, taking the i-th preset point as the center.

[0080] S2023: If the rectangular area does not exceed the boundary of the j-th image frame, extract image data corresponding to the rectangular area in the j-th image frame to obtain a slice image corresponding to the i-th preset point.

[0081] S2024: If the rectangular area exceeds the boundary of the image frame, image data corresponding to the rectangular area is extracted from the j-th image frame and adjacent image frames to obtain a slice image corresponding to the i-th preset point.

[0082] Multiple image frames are scanned in an ideal arrangement direction and have corresponding frame numbers according to the scanning sequence. Image data of the image frames, including image width, height, etc., can be obtained according to the scanning parameters of the scanner.

[0083] For the i-th preset point, based on the comparison between the coordinates of the preset point and the image data of a single image frame, the j-th image frame corresponding to the preset point can be determined, and then a rectangular area of ​​a fixed size is calculated with the preset point as the center. If the rectangular area is located in the i-th image frame, a slicing operation can be performed directly in the image frame to obtain the corresponding sliced ​​image. If the rectangular area exceeds the boundary of the j-th image frame, the image data corresponding to the rectangular area is extracted from the j-th image frame and its adjacent image frames respectively, and then the image data is merged to obtain the corresponding sliced ​​image. In this way, by traversing all preset points, the sliced ​​images corresponding to all preset points can be obtained.

[0084] If the ideal arrangement direction is from left to right, the adjacent image frames are determined as follows: if the rectangular area exceeds the right boundary of the image frame, the j+1th image frame is used as the adjacent image frame; if the rectangular area exceeds the left boundary of the image frame, the j-1th image frame is used as the adjacent image frame.

[0085] If the ideal arrangement direction is from bottom to top, the adjacent image frames are determined as follows: if the rectangular area exceeds the upper boundary of the image frame, the j+1th image frame is used as the adjacent image frame; if the rectangular area exceeds the lower boundary of the image frame, the j-1th image frame is used as the adjacent image frame.

[0086] In this way, it is possible to further adapt to substrate structures of different sizes and process the situation where the data of the slice image exceeds the image frame boundary, thereby improving the reliability of the calculation.

[0087] S203: Calculate the offset of the m-th slice image relative to the m-1-th slice image using a preset image matching algorithm to obtain N-1 offset values.

[0088] The image matching algorithm can be used to obtain the matching results between adjacent slice images. The matching results represent the spatial correspondence between adjacent slice images. By comparing the coordinate values ​​of the corresponding points in the matching results, the offset of the mth slice image relative to the m-1th slice image can be calculated to obtain N-1 offset values.

[0089] In a possible implementation, the image matching algorithm includes one of an image matching algorithm based on a sum of squared differences, an image matching algorithm based on normalized correlation, an image matching algorithm based on deep learning, or a phase correlation matching algorithm.

[0090] In actual measurement scenarios, you can choose according to the advantages and disadvantages of each image matching algorithm to improve the accuracy of angle measurement.

[0091] The image matching algorithm based on the sum of squared differences (SSD) method has a simple principle, fast computation, and low implementation difficulty. It can quickly find matching areas when image lighting conditions are stable, the contrast between the target and the background is clear, and there are no rotation or scaling changes. However, it is extremely sensitive to lighting changes; slight lighting differences can lead to increased matching errors; it cannot handle rotation and scaling of the target; and it is prone to mismatches if there are multiple similar areas in the image. This image matching algorithm is suitable for fast, rough matching of simple images. For example, when the approximate target location is known and the image conditions are stable, the SSD algorithm can be used to quickly screen possible matching areas before further refinement. It is also suitable for rapid retrieval of surveillance videos with extremely high real-time requirements and relatively low precision requirements, such as quickly locating an object with a fixed appearance in a video.

[0092] Normalization-related image matching algorithms are robust to changes in illumination, reducing the effects of illumination through normalization. Matching accuracy is high when the target shape is fixed and the background is relatively uniform. However, their computational complexity is relatively high, especially when the image size is large, where computational time increases significantly. They also struggle with geometric transformations such as rotation and scaling of the target. Matching is also poor when the target and background have similar grayscale. This image matching algorithm is suitable for medical imaging analysis, such as the registration of X-ray and CT images. When the image content is similar and there is a certain tolerance for illumination changes, the NCC algorithm can match images of different modalities or the same modality at different times. It can also be used for image stitching in the digitization of cultural relics. When the surface texture of the artifact is relatively fixed and the background is uniform, the NCC algorithm can effectively align multiple images.

[0093] Image matching algorithms based on deep learning possess powerful feature representation capabilities, can automatically learn complex image features, and are extremely robust to changes in lighting, rotation, scaling, and perspective. They can handle complex backgrounds and blurred images, and excel in large-scale image matching tasks. However, they require a large amount of annotated data for training, resulting in high data preparation costs. The model training process requires powerful computing resources (such as high-performance GPU clusters) and is time-consuming. The model structure is complex, the inference phase is computationally intensive, and real-time performance is poor. The algorithm also lacks interpretability, making it difficult to intuitively understand the matching decision process.

[0094] Since the shapes and structures of the N targets are consistent, the preset points are determined according to the properties of the target's arrangement on the substrate, such as arrangement direction and spacing, so that each preset point has relative consistency on the N targets. Then, the image is sliced ​​according to the preset points, so that the obtained sliced ​​images are comparable in spatial position and can focus on the local features of the target, thereby effectively avoiding the influence of the interference information of a single target itself on image matching.

[0095] The phase correlation matching algorithm is applicable to the following scenarios:

[0096] (1) For industrial online detection scenarios with high real-time requirements, such as rapid detection of angle and position deviation of parts on the production line, the phase correlation matching algorithm can complete image matching in a relatively short time.

[0097] (2) In scenarios where images are relatively stable, have low noise, and are sensitive to rotation and translation, such as the registration of satellite remote sensing images, the phase correlation matching algorithm can use its frequency domain characteristics to quickly and accurately find the translation and rotation relationship between images.

[0098] (3) For embedded device applications with limited computing resources, the algorithm has a relatively moderate computational complexity and can ensure a certain matching accuracy and efficiency under resource-constrained conditions.

[0099] In a possible implementation, a phase correlation matching algorithm is used as an example to describe in detail step S203, specifically including:

[0100] S2031: Perform Fourier transform on each of the N slice images to obtain corresponding frequency domain information.

[0101] S2032: Execute a phase correlation matching algorithm based on the frequency domain information to determine the offset value of the m-th slice image relative to the m-1-th slice image, and obtain N-1 offset values.

[0102] By performing a two-dimensional Fourier transform on each slice image, the corresponding frequency domain information can be obtained. Then, based on the phase information in the frequency domain information, a comparison is performed to determine the translation offset of the subsequent slice image relative to the current slice image. In this way, all adjacent slice images are matched to obtain N-1 offset values.

[0103] Therefore, when calculating the angle between the overall distribution position and arrangement direction of multiple chip units in the wafer, the phase correlation matching algorithm can improve the robustness to light and noise. Compared with the template matching method, it does not require manual intervention and has a higher degree of automation, thereby improving the efficiency of angle calculation.

[0104] S204: Accumulate N-1 offset values ​​to N preset points to obtain N target points.

[0105] By accumulating N-1 offset values ​​in sequence, the offset values ​​of the next N-1 preset points relative to the first preset point can be obtained. The N-1 offset values ​​accumulated in sequence are added to the coordinates of the next N-1 preset points to obtain the coordinates of the target point.

[0106] As an example, N is 3, the preset point coordinates are (0, 0), (1, 0), (2, 0), and the offset values ​​are (0, 0.5), (0, 1). Then, traverse the three preset points, and add 2 offset values ​​to the coordinates of the second and third preset points in sequence. The coordinates of the N target points are (0, 0), (1, 0.5), (2, 1.5).

[0107] The preset points are corrected based on the image matching results, so that the N target points can better represent the points with consistent relative positions on the N targets.

[0108] S205: Perform straight line fitting based on the N target points, and determine the angle between the fitted straight line and the ideal arrangement direction.

[0109] A straight line is fitted to N target points to find a line that represents their distribution trend. This line also represents the actual arrangement of the N targets. The angle between the fitted line and the arrangement direction can then be calculated based on the fitted line. When calculating the angle, the angle is calculated based on the special case of the fitted line: the actual arrangement direction is consistent with the ideal arrangement direction. If the slope of the fitted line matches the slope of the ideal arrangement direction, the angle is zero. Otherwise, the angle is calculated using slope comparison, vector method, or line equation method to quantify the deviation between the actual arrangement and the design requirements, thereby helping to evaluate the accuracy of the arrangement of multiple targets on the substrate.

[0110] In complex measurement scenarios where there is a lot of noise and outliers in the data, or in scenarios where the structure and shape of the target exhibit nonlinear characteristics, random sampling consistency algorithms, polynomial fitting methods, etc. can be used for straight line fitting.

[0111] For example, the random sampling consensus algorithm is highly robust to noise and outliers and can accurately fit straight lines in complex data; however, its computational efficiency is low, requiring multiple iterations, and the results have a certain degree of randomness.

[0112] The polynomial fitting method can fit point sets with nonlinear relationships and adapt to more complex shapes; however, it is difficult to select the polynomial order. Too high an order may lead to overfitting and affect the fitting accuracy.

[0113] The least squares method is suitable for regular angle calculation scenarios where the data is relatively clean and there are few outliers.

[0114] In a possible implementation, step S205 includes:

[0115] S2051: Fitting N target points by the least square method to obtain a fitting straight line;

[0116] S2052: Determine the angle between the fitting straight line and the ideal arrangement direction.

[0117] When calculating the angle between the overall distribution position and arrangement direction of multiple chip units in a wafer, the least squares method can quickly find the straight line that best represents the arrangement direction of the chip units, thereby improving the efficiency and stability of the angle calculation.

[0118] Therefore, through the above-mentioned angle measurement method, high-precision angle calculation between the actual arrangement direction and the ideal arrangement direction of multiple targets with consistent structures on the substrate is achieved. In addition, this angle measurement method has better stability for substrates of different sizes and different overall distribution areas of the targets in the substrate, providing solid technical support for the accuracy of the manufacturing process and the stability of product quality.

[0119] refer to Figure 3 , which is a data flow diagram of an angle measurement method provided in an embodiment of the present application, takes the ideal arrangement direction of N targets in the substrate as an example from bottom to top to illustrate the specific steps of angle calculation.

[0120] First, a computer device is used to obtain image data of the overall distribution area of ​​N targets according to their ideal arrangement direction in the substrate (the image data is determined by a two-dimensional coordinate system established with the ideal arrangement direction as the y-axis), and user input data (preset point set and image height) is received. The preset point set is the coordinates of the preset points obtained by the user on each target according to the arrangement attributes, and the image height refers to the height of a single image frame in the image data.

[0121] The preset point set, image height, and image data are then input into the slice calculation module for calculation to obtain the corresponding slice image. For each point in the preset point set, the frame number of the image frame in which it is located is calculated according to the y coordinate / image height, and its local coordinates within the corresponding image frame are determined. Then, with the local coordinates as the center, a fixed-size rectangular area is calculated. If the rectangular area is within the image frame, a fixed-size slice image is extracted from the image frame according to the rectangular area. If the rectangular area exceeds the upper boundary of the image frame, the slice information is added to the previous image frame. If it exceeds the lower boundary of the image frame, the slice information is added to the next image frame and merged with the slice information corresponding to the rectangular area in the image frame to obtain a fixed-size slice image. Then, all the preset points in the preset point set are divided into corresponding slice images and added to the image set.

[0122] If the number of slice images in the image set is greater than 1, that is, N is greater than 1, the image matching module is entered to calculate the offset values ​​between adjacent slice images through the phase correlation matching algorithm to obtain N-1 offset values.

[0123] Afterwards, the offset values ​​and the preset point set are simultaneously input into the coordinate processing module, and the N-1 offset values ​​are accumulated on the coordinates of the corresponding preset points in the preset point set to obtain the coordinates of the N target points.

[0124] Finally, the N target points are input into the angle calculation module, and a straight line is fitted to the N target points using the least squares method, and the target angle between the fitted straight line and the ideal arrangement direction is calculated.

[0125] Thus, by performing angle calculation through the above steps, the following beneficial effects are achieved:

[0126] (1) Significantly improved the accuracy of angle calculation: The application of phase correlation matching algorithm and least squares method effectively reduces image matching and straight line fitting errors, meeting the needs of high-precision measurement.

[0127] (2) Enhanced adaptability: The image slicing strategy can adapt to target structures of different sizes and shapes and operate stably in a variety of image scenarios.

[0128] (3) Achieve full automation: Automate the entire process from image input to angle calculation, improve measurement efficiency and reduce labor costs.

[0129] Based on the above embodiments, the present invention provides an angle measuring device. Figure 4 FIG. 4 is a schematic diagram of an angle measurement device provided in an embodiment of the present application. The device 400 includes a carrying unit 401, an illumination unit 402, an imaging unit 403, and a processing unit 404.

[0130] The carrying unit 401 is used to carry the object to be tested.

[0131] The lighting unit 402 is used to illuminate the object to be measured.

[0132] An imaging unit 403 is configured to acquire image data of the entire distribution area of ​​N targets on the object under test and N preset points, wherein the N preset points are located on the N targets and aligned along an ideal arrangement direction of the N targets, and the spacing between adjacent preset points in the N preset points is determined according to the spacing between adjacent targets;

[0133] The processing unit 404 includes a slice image determining unit 4041 , an offset value calculating unit 4042 , an accumulating unit 4043 , and an angle calculating unit 4044 , which are described below respectively.

[0134] The slice image determining unit 4041 is configured to determine N slice images of fixed size in the image data with N preset points as centers, respectively. The N slice images are used to identify local information of N targets.

[0135] The offset value calculation unit 4042 is used to calculate the offset value of the m-th slice image relative to the m-1-th slice image through an image matching algorithm to obtain N-1 offset values.

[0136] The accumulation unit 4043 is used to accumulate N-1 offset values ​​to N preset points to obtain N target points.

[0137] The angle calculation unit 4044 is used to perform straight line fitting based on N target points and determine the angle between the fitted straight line and the ideal arrangement direction.

[0138] As a result, high-precision angle calculation between the actual arrangement direction and the ideal arrangement direction of multiple structurally consistent targets on the substrate is achieved. In addition, it has better stability for substrates of different sizes and different overall distribution areas of the targets in the substrate, providing solid technical support for the accuracy of the manufacturing process and the stability of product quality.

[0139] In a possible implementation, the image matching algorithm includes one of an image matching algorithm based on a sum of squared differences, an image matching algorithm based on normalized correlation, an image matching algorithm based on deep learning, or a phase correlation matching algorithm.

[0140] Therefore, selecting an appropriate image matching algorithm based on the actual measurement scenario can improve the accuracy of angle measurement.

[0141] In one possible implementation, the offset calculation unit is specifically used to: perform Fourier transform on N slice images respectively to obtain corresponding frequency domain information; perform matching based on the frequency domain information to determine the offset value of the mth slice image relative to the m-1th slice image to obtain N-1 offset values.

[0142] Therefore, when calculating the angle between the overall distribution position and arrangement direction of multiple chip units in the wafer, the phase correlation matching algorithm can improve the robustness to light and noise. Compared with the template matching method, it does not require manual intervention and has a higher degree of automation, thereby improving the efficiency of angle calculation.

[0143] In one possible implementation, the image data includes multiple image frames scanned in an ideal arrangement direction, and the slice image determination unit is specifically used to: determine the jth image frame corresponding to the i-th preset point for the i-th preset point; calculate a rectangular area according to a fixed size with the i-th preset point as the center; if the rectangular area does not exceed the boundary of the j-th image frame, extract the image data corresponding to the rectangular area in the j-th image frame to obtain the slice image corresponding to the i-th preset point; if the rectangular area exceeds the boundary of the image frame, extract the image data corresponding to the rectangular area in the j-th image frame and the adjacent image frames to obtain the slice image corresponding to the i-th preset point.

[0144] In this way, it is possible to further adapt to substrate structures of different sizes and process the situation where the data of the slice image exceeds the image frame boundary, thereby improving the reliability of the calculation.

[0145] In a possible implementation, the angle calculation unit is specifically configured to: fit the N target points by a least square method to obtain a fitting straight line; and determine an angle between the fitting straight line and the ideal arrangement direction.

[0146] Therefore, when calculating the angle between the overall distribution position and arrangement direction of multiple chip units in the wafer, the least squares method can quickly find the straight line that best represents the arrangement direction of the chip units, thereby improving the efficiency and stability of the angle calculation.

[0147] In one possible implementation, the target is a chip unit and the substrate is a wafer. It can be understood that according to the present application solution, the alignment accuracy during the wafer manufacturing process can be evaluated and optimized, thereby improving the overall quality and performance of the wafer.

[0148] Based on the above embodiments, an embodiment of the present application provides a computer-readable storage medium, which is used to store a computer program. When the computer program is executed by a computer device, it implements the above angle measurement method.

[0149] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems or devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0150] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An angle measurement method, characterized in that: The method comprises: Acquire image data of an overall distribution area of ​​N targets in a substrate and N preset points, wherein the N preset points are respectively located on the N targets and aligned along an ideal arrangement direction of the N targets, and the spacing between adjacent preset points in the N preset points is determined according to the spacing between adjacent targets; Determining N slice images of fixed sizes in the image data with the N preset points as centers, respectively, the N slice images being used to identify local information of the N targets; Calculate the offset value of the mth slice image relative to the m-1th slice image using a preset image matching algorithm to obtain N-1 offset values; Accumulate the N-1 offset values ​​to the N preset points to obtain N target points; A straight line fitting is performed based on the N target points, and an angle between the fitting straight line and the ideal arrangement direction is determined.

2. The method according to claim 1, characterized in that Calculating the offset value of the mth slice image relative to the m-1th slice image to obtain N-1 offset values ​​includes: Performing discrete Fourier transform on the N slice images respectively to obtain corresponding frequency domain information; A phase correlation matching algorithm is executed based on the frequency domain information to determine the offset value of the m-th slice image relative to the m-1-th slice image, thereby obtaining N-1 offset values.

3. The method according to claim 1, characterized in that The image matching algorithm includes any one of an image matching algorithm based on the sum of squared differences, an image matching algorithm based on normalized correlation, an image matching algorithm based on deep learning, or a phase correlation matching algorithm.

4. The method according to claim 1, wherein The image data includes a plurality of image frames scanned according to the ideal arrangement direction, and the N slice images of fixed sizes determined in the image data with the N preset points as centers respectively include: For the i-th preset point, determining the j-th image frame corresponding to the i-th preset point; Calculate a rectangular area with a fixed size, taking the i-th preset point as the center; If the rectangular area does not exceed the boundary of the j-th image frame, extracting image data corresponding to the rectangular area in the j-th image frame to obtain a slice image corresponding to the i-th preset point; If the rectangular area exceeds the boundary of the image frame, image data corresponding to the rectangular area is extracted from the j-th image frame and adjacent image frames to obtain a slice image corresponding to the i-th preset point.

5. The method according to claim 1, wherein The performing straight line fitting based on the N target points and determining the angle between the fitting straight line and the ideal arrangement direction includes: Fitting the N target points using the least squares method to obtain a fitting straight line; The angle between the fitting straight line and the ideal arrangement direction is determined.

6. The method according to claim 1, characterized in that The target is a chip unit, and the substrate is a wafer.

7. A semiconductor measuring device, characterized in that: The device includes a carrying unit, an illumination unit, an imaging unit, and a processing unit: The carrying unit is used to carry the object to be tested; The lighting unit is used to illuminate the object to be tested; The imaging unit is used to obtain image data of the entire distribution area of ​​N targets on the object to be measured in the substrate and N preset points, wherein the N preset points are respectively located on the N targets and aligned along the ideal arrangement direction of the N targets, and the spacing between adjacent preset points in the N preset points is determined according to the spacing between adjacent targets; The processing unit includes a slice image determination unit, an offset value calculation unit, an accumulation unit, and an angle calculation unit: The slice image determination unit is used to determine N slice images of fixed size in the image data with the N preset points as centers, respectively, and the N slice images are used to identify local information of the N targets; The offset value calculation unit is used to calculate the offset value of the mth slice image relative to the m-1th slice image by a preset image matching algorithm to obtain N-1 offset values; The accumulating unit is configured to accumulate the N-1 offset values ​​to the N preset points to obtain N target points; The angle calculation unit is used to perform straight line fitting based on the N target points and determine the angle between the fitting straight line and the ideal arrangement direction.

8. The device according to claim 7, characterized in that The offset value calculation unit is specifically used for: Performing Fourier transform on the N slice images respectively to obtain corresponding frequency domain information; A phase correlation matching algorithm is executed based on the frequency domain information to determine the offset value of the m-th slice image relative to the m-1-th slice image, thereby obtaining N-1 offset values.

9. The device according to claim 7, characterized in that The image data includes a plurality of image frames scanned according to the ideal arrangement direction, and the image determination unit is specifically configured to: For the i-th preset point, determining the j-th image frame corresponding to the i-th preset point; Calculate a rectangular area with a fixed size, taking the i-th preset point as the center; If the rectangular area does not exceed the boundary of the j-th image frame, extracting image data corresponding to the rectangular area in the j-th image frame to obtain a slice image corresponding to the i-th preset point; If the rectangular area exceeds the boundary of the image frame, image data corresponding to the rectangular area is extracted from the j-th image frame and adjacent image frames to obtain a slice image corresponding to the i-th preset point.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store a computer program, and when the computer program is executed by a computer device, the computer program implements the method according to any one of claims 1 to 6.

Images