Method and apparatus for measuring semiconductor microstructure dimensions

Abstract

The application discloses a semiconductor microstructure size measurement method and device, electronic equipment and storage medium. The method comprises the following steps: adopting a general semiconductor microstructure size measurement method to perform size measurement on an obtained semiconductor image, and obtaining a microstructure size measurement result; wherein the general semiconductor microstructure size measurement method adopts an open set image segmentation model to perform image segmentation on the semiconductor image, and the open set image segmentation model supports segmentation of any semiconductor microstructure; the microstructure size measurement result is input into a closed set image segmentation model as a training data set for model training, so as to adopt a special semiconductor microstructure size measurement method to perform size measurement on a to-be-measured semiconductor microstructure; wherein the closed set image segmentation model is used for segmentation of the semiconductor microstructure included in the training data set. By adopting the scheme, the microstructure size measurement is automatically completed, the measurement time is shortened, the artificial cost is reduced, and the work efficiency of an analyst is improved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, electronic device, and storage medium for measuring the dimensions of semiconductor microstructures. Background Technology

[0002] In fields such as semiconductor failure analysis and materials analysis, analysts often need to measure the size of the microstructural units (such as grains and particles) of materials under specific scenarios.

[0003] In traditional analytical workflows, analysts employ manual measurement methods. First, they use an electron microscope to image the microstructure. Then, using appropriate software, analysts visually identify each microstructural unit (such as a grain or particle) in the image and select (click) the range of dimensions to be measured within that unit (e.g., the two endpoints of the major axis) using a mouse. Afterward, they record the readings displayed by the software. To generate relevant reports, analysts also manually plot the measurement dimensions and other information onto the image and use office software such as Excel to record and compile data.

[0004] Because analysts need to manually measure dozens to hundreds of microstructural units in each image, this is not only time-consuming and labor-intensive (usually taking several hours), but the repetitive and tedious work can also easily lead to mental fatigue, thus affecting the accuracy and timeliness of the analysis data. Furthermore, different analysts may use different measurement standards at different times, and even if they share the same understanding of the measurement standards, errors can still occur during actual measurements due to mouse clicks, leading to issues with the consistency and repeatability of the analysis data. The mental fatigue resulting from prolonged measurements further exacerbates the errors and repeatability problems in data analysis. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for measuring the dimensions of semiconductor microstructures, in order to solve the problems of high time consumption, high cost, and low accuracy of manual measurement.

[0006] According to one aspect of the present invention, a method for measuring the dimensions of semiconductor microstructures is provided, the method comprising:

[0007] A general semiconductor microstructure size measurement method is used to measure the size of the acquired semiconductor image to obtain the microstructure size measurement results. Among them, the general semiconductor microstructure size measurement method uses an open set image segmentation model to segment the semiconductor image. The open set image segmentation model supports segmentation of arbitrary semiconductor microstructures.

[0008] The microstructure size measurement results are used as a training dataset and input into a closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method; wherein, the closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset.

[0009] According to another aspect of the present invention, a semiconductor microstructure size measuring device is provided, the device comprising:

[0010] A general semiconductor microstructure size measurement module is used to measure the size of acquired semiconductor images using a general semiconductor microstructure size measurement method, and obtain the microstructure size measurement results. The general semiconductor microstructure size measurement method uses an open set image segmentation model to segment the semiconductor image, and the open set image segmentation model supports segmentation of arbitrary semiconductor microstructures.

[0011] A dedicated semiconductor microstructure size measurement module is used to input the microstructure size measurement results as a training dataset into a closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method; wherein, the closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset.

[0012] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0013] At least one processor; and

[0014] A memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the semiconductor microstructure size measurement method according to any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the semiconductor microstructure size measurement method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the semiconductor microstructure size measurement method as described in any embodiment of the present invention.

[0018] The technical solution of this invention measures the size of a semiconductor image using a general semiconductor microstructure size measurement method, obtaining the microstructure size measurement result. This general method employs an open-set image segmentation model to segment the semiconductor image, supporting the segmentation of any semiconductor microstructure. The microstructure size measurement result is then used as a training dataset and input into a closed-set image segmentation model for training, enabling the measurement of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method. The closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset. This solution solves the problems of high time consumption, low accuracy, and high cost associated with manual measurement, achieving automated microstructure size measurement, reducing measurement time to a few minutes, minimizing manual intervention costs, and significantly improving the efficiency of analysts.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a semiconductor microstructure size measurement method provided in Embodiment 1 of the present invention;

[0022] Figure 2 This is a flowchart of another semiconductor microstructure size measurement method provided in Embodiment 1 of the present invention;

[0023] Figure 3 This is a flowchart of a semiconductor microstructure size measurement method according to Embodiment 2 of the present invention;

[0024] Figure 4 This is a schematic diagram illustrating the determination of the image format conversion scale position according to Embodiment 2 of the present invention;

[0025] Figure 5 This is a schematic diagram of a scale detection algorithm provided in Embodiment 2 of the present invention;

[0026] Figure 6 This is a schematic diagram of a semiconductor image obtained from a different angle according to Embodiment 2 of the present invention;

[0027] Figure 7 This is a schematic diagram of a semiconductor image obtained from another different angle according to Embodiment 2 of the present invention;

[0028] Figure 8 This is a flowchart of a semiconductor microstructure size measurement method according to Embodiment 3 of the present invention;

[0029] Figure 9 This is a schematic diagram of a semiconductor microstructure size measuring device according to Embodiment 4 of the present invention;

[0030] Figure 10 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0032] The acquisition, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. It should be noted that the terms "first," "second," "target," and "original," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising," "etc.," and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] Example 1

[0034] Figure 1This is a flowchart of a semiconductor microstructure size measurement method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where a general semiconductor microstructure size measurement method is first used to measure the microstructure of an open-set image segmentation model, and the microstructure size measurement result is used as the dataset of a closed-set image segmentation model. Then, a dedicated semiconductor microstructure size measurement method is used to measure the size of the semiconductor microstructure to be measured, thereby achieving automatic measurement. This method can be executed by a semiconductor microstructure size measurement device, which can be implemented in hardware and / or software. This semiconductor microstructure size measurement device can be configured in any electronic device with network communication capabilities. Figure 1 As shown, the method includes:

[0035] S110. The size of the acquired semiconductor image is measured using a general semiconductor microstructure size measurement method to obtain the microstructure size measurement results.

[0036] The general semiconductor microstructure size measurement method uses an open-set image segmentation model to segment semiconductor images, such as the SAM model. This open-set image segmentation model supports segmentation of arbitrary semiconductor microstructures. This general semiconductor microstructure size measurement method can easily adapt to different types and shapes of microstructure units, thus making it a universal measurement method.

[0037] In this embodiment of the invention, a general semiconductor microstructure size measurement method is used to measure the size of the acquired semiconductor image in an open set image segmentation model to obtain the microstructure size measurement results.

[0038] S120. The microstructure size measurement results are used as a training dataset and input into the closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method.

[0039] Among these methods, a dedicated semiconductor microstructure size measurement approach utilizes closed-set image segmentation models, such as the Unet model and the Deeplab model. These closed-set image segmentation models are used to segment semiconductor microstructures included in the training dataset. Since closed-set image segmentation models cannot accommodate microstructure units outside the training data, this is a dedicated measurement method.

[0040] General-purpose semiconductor microstructure size measurement methods require complex post-processing logic to extract microstructural units. Specialized semiconductor microstructure size measurement methods, on the other hand, do not require complex post-processing; the segmented objects can be directly used for measurement. Closed-set image segmentation models segment microstructural units more accurately than open-set image segmentation models, thus representing a high-precision method. However, due to the difficulty in obtaining training data and the high cost of annotation for closed-set image segmentation models, [see...]. Figure 2 In the implementation of this invention, an open-set image segmentation model is first used to measure the size of the microstructure, and the measurement results of the microstructure size are used as the training dataset for the closed-set image segmentation model to train a dedicated high-precision closed-set image segmentation model for this type of microstructure offline. The measurement results obtained by using the dedicated model are more accurate, and the number of times the user needs to intervene manually is less than that of the general model.

[0041] In one optional embodiment of the present invention, each closed-set image segmentation model can only be used to segment a specific type of microstructure, i.e., the types of microstructures contained in the training data. Therefore, to adapt to the size measurement of multiple microstructures, it may be necessary to train multiple closed-set image segmentation models. For example, an analyst initially measures a batch of images using a general semiconductor microstructure size measurement method, and this batch of images contains only one type of microstructure unit; then the analyst uses the result data to train a dedicated segmentation model, which can only be used for that type of microstructure unit. If the analyst subsequently wants to measure another type of microstructure unit, then the analyst needs to measure a batch of images again using the general semiconductor microstructure size measurement method, and then train a dedicated model. However, since the input of the closed-set image segmentation model is the measurement result obtained using the general semiconductor microstructure size measurement method, no time is wasted.

[0042] This invention provides a method for measuring the dimensions of semiconductor microstructures. By employing a general semiconductor microstructure dimension measurement method, the method measures the dimensions of an acquired semiconductor image using an open-set image segmentation model, obtaining the microstructure dimension measurement results. The open-set image segmentation model supports segmenting any semiconductor microstructure. The microstructure dimension measurement results are used as a training dataset and input into a closed-set image segmentation model for model training, enabling the measurement of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure dimension measurement method. The closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset. This invention automatically completes the microstructure dimension measurement process, reducing measurement time to a few minutes, minimizing manual intervention costs, and significantly improving the efficiency of analysts.

[0043] Example 2

[0044] Figure 3 This is a flowchart of a semiconductor microstructure size measurement method provided in Embodiment 2 of the present invention. The embodiments of the present invention further optimize the aforementioned embodiments, and can be combined with various optional solutions from one or more of the above embodiments. For example... Figure 3 As shown, the method includes:

[0045] S310. Convert the acquired semiconductor image into an image format and determine the pixel distance conversion relationship of the semiconductor image.

[0046] Since image analysis algorithms can only process common computer images, such as PNG or JPEG, measurements on these images can only obtain pixel-level results. However, the microstructural size measurements required in this embodiment of the invention need to be measured in actual physical dimensions, in units of nanometers or micrometers. Therefore, a conversion relationship is needed to convert units from pixels to nanometers or micrometers, called the pixel distance conversion relationship.

[0047] The semiconductor image refers to a microscopic image acquired using an electron microscope. Microscopic images differ from computer-acquired images and require special software to open. Before determining the pixel distance conversion relationship of the semiconductor image, the microscopic image needs to be converted to a different image format. This conversion can involve converting the microscopic image to a common computer image format, such as PNG or JPEG. The reading and format conversion of the microscopic image can be performed using existing tools, and no specific limitations are imposed in this embodiment of the invention.

[0048] As an optional but non-limiting implementation, the step of converting the acquired semiconductor image into an image format and determining the pixel distance conversion relationship of the semiconductor image includes, but is not limited to, steps A1-A2:

[0049] Step A1: Convert the semiconductor image obtained by electron microscopy into a semiconductor image in a common computer format; wherein the semiconductor image in the common computer format includes a scale bar.

[0050] Step A2: Identify and detect the text portion of the scale bar and the bar diagram portion representing length to determine the pixel distance conversion relationship of the semiconductor image; wherein, the pixel distance conversion relationship is the ratio of the value of the text portion of the scale bar to the pixel value of the bar diagram portion.

[0051] This invention involves converting semiconductor images acquired using an electron microscope into a standard computer-formatted semiconductor image. For standard computer images, this invention requires a scale bar in the lower left corner of the image for algorithm extraction. Generally, scale bars can be included with images taken using an electron microscope during the conversion process, such as... Figure 4 As shown.

[0052] Optionally, the acquired semiconductor image can also refer to a computer image in a common format. For computer images in a common format, no format conversion is required, and the scale bar contained in the computer image can be directly determined. By detecting and recognizing the text portion of the scale bar and the bar diagrams used to represent length, the pixel distance conversion relationship can be determined.

[0053] The scale bar includes a text section and bar charts representing lengths. The length of the bar charts (in pixels) is equivalent to the distance shown in the text section. For example... Figure 4 The scale shown indicates that the length of the white bar diagram is equivalent to 200 nanometers. Therefore, the distance conversion relationship T can be calculated from this. Let the length of the bar diagram be L pixels, the text value be N, and the unit be U, then the distance conversion relationship... (U per pixel), where U can be nanometers or micrometers. The distance conversion can be calculated as long as the text and bar chart portions are successfully detected. If detection fails, the subsequent algorithm will still output measurement results in pixels. Analysts can perform the conversion themselves.

[0054] Among them, see Figure 5 The text area is cropped from the original image, typically the lower left corner. An Optical Character Recognition (OCR) model is used to identify the text in the cropped area. Any open-source OCR model that meets the requirements can be used, such as the Paddle OCR model; this invention does not impose any restrictions. The numerical value N and unit U of the text are extracted based on the OCR results. If the numerical value N and unit U are detected successfully, proceed to the next step; otherwise, exit. A strip pattern detection area is further cropped from the cropped area for strip pattern detection. Since microscopic images are black and white, the strip pattern is usually a white rectangular strip with a black border; the white portion is the area to be detected. First, the image is binarized to extract the white area; second, rectangular objects are extracted from the binary image based on their contours; finally, the strip with the largest area among the rectangular objects is the final detected strip. If a rectangular strip is detected, its length (rightmost coordinate minus leftmost coordinate) is calculated, and the distance transformation relationship is calculated. Otherwise, exit.

[0055] In this embodiment of the invention, the pixel distance conversion relationship of the acquired semiconductor image is determined so that the microstructure size can be measured subsequently.

[0056] S320. Use an open set image segmentation model to segment the semiconductor image and determine the image segmentation result.

[0057] Image segmentation refers to the technique and process of dividing an image into several specific regions with unique properties and identifying targets of interest. In this embodiment of the invention, image segmentation is performed on a semiconductor image to determine the segmentation result. The image segmentation result includes at least two masks, each mask comprising at least one semiconductor microstructure. Optionally, the image segmentation method is not specifically limited in this embodiment of the invention.

[0058] S330. Preprocess at least two masks included in the image segmentation result to obtain a target mask set.

[0059] Preprocessing can refer to processing at least two masks included in the image segmentation result to obtain a clear target mask containing only one semiconductor microstructure. The preprocessing includes, but is not limited to, removing overlapping masks, removing noise from masks, removing noisy masks, removing masks located at image edges, removing masks in the image scale area, and removing background masks.

[0060] As an optional but non-limiting implementation, the image segmentation result includes preprocessing of at least two masks to obtain a target mask set, including but not limited to steps B1-B6:

[0061] Step B1: Perform overlapping mask removal processing on at least two masks included in the image segmentation result to obtain a first mask set; wherein, the overlapping mask removal processing includes obtaining overlapping masks, determining the area size of overlapping masks, and removing overlapping masks with smaller areas.

[0062] Step B2: Using image connected component analysis, noise in the first mask set is removed to obtain the second mask set; wherein each mask in the second mask set includes only one semiconductor microstructure.

[0063] Step B3: Remove the noise masks in the second mask set to obtain the third mask set.

[0064] Step B4: Remove the masks located at the edge of the semiconductor image in the third mask set to obtain the fourth mask set.

[0065] Step B5: Remove the masks located in the scale region from the fourth mask set to obtain the fifth mask set.

[0066] Step B6: Remove the masks belonging to the background region in the fifth mask set to obtain the target mask set.

[0067] The process involves determining whether there are partially overlapping masks among at least two masks in the image segmentation result. This requires removing the area occupied by the smaller mask from the larger mask to ensure no overlap between masks, resulting in a first mask set. Image connected component analysis is then applied to the first mask set to remove noise from each mask, ensuring only one semiconductor microstructure per mask, resulting in a second mask set. Some potentially noisy, small-area masks are removed from the second mask set, resulting in a third mask set. Masks located at the image edges are removed from the third mask set, as parts of objects at the image edges are outside the image and cannot be measured; therefore, these parts need to be removed, resulting in a fourth mask set. Masks in the scale region are removed from the fourth mask set, resulting in a fifth mask set. If the image contains a scale bar, the nearby microstructure units are occluded and cannot be accurately measured; therefore, the corresponding mask also needs to be removed. Background objects are removed from the fifth mask set to obtain the final target mask set.

[0068] As an optional but non-limiting implementation, removing the masks belonging to the background region from the fifth mask set to obtain the target mask set includes, but is not limited to, steps C1-C3:

[0069] Step C1: Determine the pixel grayscale value of the semiconductor image, and determine a first threshold and a second threshold based on the pixel grayscale value; wherein, the first threshold is greater than the second threshold, the first threshold is used to distinguish the brighter areas in the semiconductor image, and the second threshold is used to distinguish the darker areas in the semiconductor image.

[0070] Step C2: Determine the first mean and median of the image grayscale values ​​corresponding to the fifth mask; if the median of the image grayscale values ​​of the fifth mask is greater than the first threshold, then the fifth mask is determined to be a bright mask; if the median of the image grayscale values ​​of the fifth mask is less than the second threshold, then the fifth mask is determined to be a dark mask; if the median of the image grayscale values ​​of the fifth mask is between the first threshold and the second threshold, then determine the second mean of the image grayscale values ​​within a preset range outside the fifth mask; if the first mean is greater than the second mean, then the fifth mask is a bright mask, otherwise it is a dark mask; wherein, the set of fifth masks includes at least two fifth masks.

[0071] Step C3: Determine the number of bright masks and dark masks in the fifth mask set. If the number of bright masks is greater than the number of dark masks, then the bright masks are used as the target mask set; if the number of bright masks is less than the number of dark masks, then the dark masks are used as the target mask set.

[0072] The open-set image segmentation model not only segments possible microstructural units but may also segment some background regions, which need to be removed. Microscopic images are divided into bright-field and dark-field images. Bright-field images are obtained by direct imaging with transmitted electrons, and typically have a bright background; dark-field images are obtained by imaging with diffracted electrons, and typically have a dark background. Based on this characteristic, embodiments of the present invention can determine which parts are foreground and which are background based on the number of bright and dark masks. The specific steps are as follows:

[0073] 1) Dynamically determine the thresholds for bright and dark areas in the image (this varies for each image). Statistically analyze the pixel grayscale values ​​of the semiconductor image and determine a first and second threshold based on these values. For example, the 90th percentile pixel grayscale value is used as the first threshold for bright areas (greater than this threshold), and the 40th percentile pixel grayscale value is used as the second threshold for dark areas (less than this threshold). Note that 90% and 40% are empirical values ​​and need to be adjusted based on actual conditions.

[0074] 2) For each fifth mask, calculate the mean and median of the grayscale values ​​of the corresponding image pixels. If the median is greater than the first threshold for the brighter areas, the fifth mask is considered bright; if the median is less than the second threshold for the darker areas, the fifth mask is considered dark. If the median is between the first threshold for the brighter areas and the second threshold for the darker areas, the judgment is made based on the brightness contrast between the pixels inside and around the fifth mask. Specifically, compare the mean grayscale value of each pixel inside the fifth mask with the mean grayscale value of pixels within a certain range around the fifth mask. If the former is greater than the latter, the fifth mask is considered bright; otherwise, it is considered dark.

[0075] 3) Determine the number of bright and dark masks. If the number of bright masks is greater than the number of dark masks, the background is considered dark, and only the bright masks are used as the target masks. Conversely, if the background is considered bright, only the dark masks are used as the target masks.

[0076] In this embodiment of the invention, the mask obtained by image segmentation is processed to determine the target mask for which microscopic result size measurement is required.

[0077] S340. Determine the size measurement result of the target mask based on the pixel distance conversion relationship of the semiconductor image; wherein, the target mask includes a semiconductor microstructure.

[0078] Specifically, after determining the target mask, the size measurement result of the target mask is determined according to a pre-determined pixel distance transformation relationship. For example, after determining the target mask, the area of ​​the target mask in the image is determined based on the pixels of the target mask, and then the size of the semiconductor microstructure included in the target mask is determined according to the pixel distance transformation relationship.

[0079] As an optional but non-limiting implementation, determining the size measurement result of the target mask based on the pixel distance transformation relationship of the semiconductor image includes, but is not limited to, steps D1-D3:

[0080] Step D1: Determine the first and second moments of the target mask, and determine the principal and secondary axes of the target mask based on the first and second moments.

[0081] Step D2: Determine the first pixel distance between the intersection of the main axis of the target mask and the outline of the target mask, and determine the actual physical major axis of the target mask based on the first pixel distance and the pixel distance conversion relationship.

[0082] Step D3: Determine the second pixel distance between the intersection point of the target mask sub-axis and the target mask contour, and determine the actual physical minor axis of the target mask based on the second pixel distance and the pixel distance conversion relationship.

[0083] The process involves determining the first and second moments of the target mask. The centroid coordinates of the target mask are calculated based on the first moment, and the covariance matrix of the target mask is calculated based on the second moment. Furthermore, eigenvalue decomposition is performed on the covariance matrix to obtain the representations of the principal and secondary axes. The intersection points of the principal axis and the target mask contour are calculated, and the distance between the two intersection points is determined as the major axis of the target mask. The intersection points of the secondary axis and the contour are calculated, and the distance between the two intersection points is determined as the minor axis of the target mask. After determining the major and minor axes of the target mask in the image, the actual physical major and minor axes of the target mask are determined based on pixel distance transformation relationships.

[0084] Optionally, since the microstructural units are irregularly shaped and their radii are not explicitly defined, this embodiment of the invention constructs three methods for calculating the radius, including:

[0085] The first method involves calculating the minimum circumcircle of the target mask and determining its radius; then, based on the pixel distance conversion relationship, determining the actual physical radius of the target mask.

[0086] The second method involves calculating the length of the target mask contour and using that contour length as the circumference of a circle. The radius of the circle is then calculated using the circumference formula, and the actual physical radius of the target mask is determined based on the pixel distance conversion relationship.

[0087] The third method involves determining the area of ​​the target mask based on its pixels, using that area as the area of ​​a circle, calculating its radius using the formula for the area of ​​a circle, and determining the actual physical radius of the target mask based on the pixel distance conversion relationship.

[0088] In this embodiment of the invention, the size of the target mask is measured according to a predetermined pixel distance conversion relationship in order to determine the size measurement results of the microstructure.

[0089] As an optional but non-limiting implementation, before determining the size measurement result of the target mask based on the pixel distance transformation relationship of the semiconductor image, the method may include, but is not limited to, steps E1-E3:

[0090] Step E1: If there are at least two sixth masks in the target mask that are obtained from the same position but different angles, then determine the minimum bounding rectangle of the at least two sixth masks; the different angles include the shooting angle and the electron diffraction angle.

[0091] Step E2: If the first cross-union ratio of the minimum bounding rectangles of the at least two sixth masks is greater than the first cross-union ratio threshold, then determine the second cross-union ratio of the at least two sixth masks; if the second cross-union ratio is greater than the second cross-union ratio threshold, then retain the sixth mask with higher confidence.

[0092] Step E3: If the first intersection-union ratio of the minimum bounding rectangle of the at least two sixth masks is less than the first intersection-union ratio threshold, then the at least two sixth masks are retained simultaneously.

[0093] In one optional embodiment of the present invention, semiconductor images may be acquired at the same location but from different angles. For example, in some dark-field measurements using transmission electron microscopy, by adjusting the objective aperture of the transmission electron microscope, different diffracted electron beams can be captured, obtaining multiple diffraction images of the photographed object. In these multiple images acquired in this manner, several identical and different microstructural units can be observed; such as... Figure 6 as well as Figure 7 The image shown is a semiconductor image acquired from different angles. In this case, since it is not possible to repeat measurements on microstructure units that have already been measured, it is necessary to perform deduplication processing on masks belonging to the same location but at different angles.

[0094] Specifically, this invention presents a non-maximum suppression (NMS) algorithm suitable for masks. The core logic of this algorithm is to first calculate the intersection-union ratio (IU) between the bounding rectangles of two masks. If the IU is greater than a certain threshold (e.g., 0.8), then the IU between the two masks is directly calculated. If the IU between the two masks is still greater than a certain threshold (e.g., 0.8), the mask with the higher confidence level is retained, and the other mask is discarded. The confidence level comes from the output of the image segmentation model. It should be noted that calculating the IU between the bounding rectangles is not mandatory, but including this step can significantly improve the algorithm's running speed. After comparing the overlap between all masks, the deduplicated result can be obtained.

[0095] As an optional but non-limiting implementation, after using a general semiconductor microstructure size measurement method to measure the size of the acquired semiconductor image in an open-set image segmentation model and obtaining the microstructure size measurement result, the method further includes:

[0096] A display interface for microstructure size measurement results is constructed, and microstructure size measurement results with errors are corrected.

[0097] In this embodiment of the invention, a microstructure size measurement result display interface is also constructed to allow users to correct the microstructure size measurement results. Simultaneously, the microstructure size measurement result display interface can also be used to label the microstructure size measurement data and use the labeled data as the training dataset for the closed-set image segmentation model used in the dedicated semiconductor microstructure size measurement method.

[0098] This invention provides a method for measuring the size of semiconductor microstructures. The method involves converting the acquired semiconductor image into an image format and determining the pixel distance transformation relationship of the semiconductor image; segmenting the semiconductor image using an open-set image segmentation model to determine the image segmentation result; wherein the image segmentation result includes at least two masks, each mask including at least one semiconductor microstructure; preprocessing the at least two masks included in the image segmentation result to obtain a target mask set; wherein the preprocessing includes removing overlapping masks, removing noise from the masks, removing noise masks, removing masks located at image edges, removing masks in the image scale area, and removing background masks; and determining the size measurement result of the target masks based on the pixel distance transformation relationship of the semiconductor image. Using the technical solution of this invention, the size measurement of microstructures is automatically completed using a general semiconductor microstructure size measurement method, reducing labor costs and improving the accuracy and efficiency of size measurement; simultaneously, the microstructure size measurement result is used as training data for a closed-set image segmentation model, solving the problem of difficulty in obtaining training data.

[0099] Example 3

[0100] Figure 8 This is a flowchart of a semiconductor microstructure size measurement method provided in Embodiment 3 of the present invention. The embodiments of the present invention further optimize the aforementioned embodiments, and can be combined with various optional solutions from one or more of the above embodiments. For example... Figure 8 As shown, the method includes:

[0101] S810. Use the microstructure size measurement results as a training dataset.

[0102] In this invention, the microstructure size measurement results determined using a general semiconductor microstructure size measurement method are used as the training dataset for a dedicated semiconductor microstructure size measurement method to address the difficulty in obtaining training data for closed-set image segmentation models. Furthermore, the microstructure size measurement results determined using the general semiconductor microstructure size measurement method in this embodiment can be considered correct, thus requiring no further annotation and saving labor costs.

[0103] S820. Input the training dataset into the closed-set image segmentation model for model training to obtain a converged target closed-set image segmentation model.

[0104] In this embodiment, a training dataset is input into a closed-set image segmentation model to train the model and obtain a converged target closed-set image segmentation model. The method for training the model and determining the convergence of the closed-set image segmentation model are not specifically limited in this embodiment.

[0105] S830. A dedicated semiconductor microstructure size measurement method is used to measure the size of the microstructure of the semiconductor to be measured in the target closed set image segmentation model.

[0106] In this process, after the target closed set image segmentation model is trained, a dedicated semiconductor microstructure size measurement method is used to measure the size of the microstructure of the semiconductor to be measured.

[0107] As an optional but non-limiting implementation, the method employs a dedicated semiconductor microstructure size measurement method to measure the size of the semiconductor to be measured within the target closed-set image segmentation model, including but not limited to steps F1-F3:

[0108] Step F1: Convert the acquired semiconductor image to be measured into an image format and determine the pixel distance conversion relationship of the semiconductor image to be measured.

[0109] Step F2: Use the target closed set image segmentation model to segment the image of the semiconductor to be measured and determine the image segmentation result; wherein, the image segmentation result includes at least two target masks, and each target mask includes a semiconductor microstructure to be measured.

[0110] Step F3: Determine the size measurement result of the target mask based on the pixel distance conversion relationship of the semiconductor image to be measured.

[0111] The method for measuring the microstructure dimensions of a dedicated semiconductor is similar to that for measuring the microstructure dimensions of a general semiconductor. Since the dedicated method uses a closed-set image segmentation model for measurement, the objects being measured only include those included in the training data. Therefore, there is no need to process the mask obtained after image segmentation.

[0112] Specifically, the acquired image of the semiconductor to be measured is converted to a different image format, and the pixel distance transformation relationship of the image is determined. A target closed-set image segmentation model is used to segment the image of the semiconductor to be measured, and the image segmentation result is determined. The image segmentation result includes at least two target masks, each target mask including a microstructure of the semiconductor to be measured. Based on the pixel distance transformation relationship of the image of the semiconductor to be measured, the size measurement result of the target masks is determined.

[0113] In this embodiment of the invention, the microstructure size measurement results obtained using a general semiconductor microstructure size measurement method are used as the training dataset. The trained high-precision dedicated closed-set image segmentation model can further improve measurement accuracy and reduce manual intervention. Simultaneously, a dedicated semiconductor microstructure size measurement method is used to automatically measure the size of the semiconductor microstructure to be measured. The automatic measurement standard is uniform, avoiding the problem of inconsistent measurement standards that may arise from differences in understanding and fatigue during manual measurement.

[0114] Example 4

[0115] Figure 9 This is a schematic diagram of a semiconductor microstructure size measuring device provided in Embodiment 4 of the present invention. Figure 9 As shown, the device includes:

[0116] The general semiconductor microstructure size measurement module 910 is used to measure the size of the acquired semiconductor image using a general semiconductor microstructure size measurement method, and obtain the microstructure size measurement result; wherein, the general semiconductor microstructure size measurement method uses an open set image segmentation model to segment the semiconductor image, and the open set image segmentation model supports segmentation of arbitrary semiconductor microstructures;

[0117] A dedicated semiconductor microstructure size measurement module 920 is used to input the microstructure size measurement results as a training dataset into a closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method; wherein, the closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset.

[0118] Optionally, the general-purpose semiconductor microstructure size measurement module is also specifically used for:

[0119] The acquired semiconductor image is converted to an image format, and the pixel distance conversion relationship of the semiconductor image is determined.

[0120] An open-set image segmentation model is used to segment a semiconductor image and determine the image segmentation result; wherein the image segmentation result includes at least two masks, and each mask includes at least one semiconductor microstructure;

[0121] The image segmentation result includes at least two masks, which are preprocessed to obtain a target mask set; wherein the preprocessing includes removing overlapping masks, removing noise from the masks, removing noise masks, removing masks located at the image edges, removing masks in the image scale area, and removing background masks.

[0122] The size measurement result of the target mask is determined based on the pixel distance conversion relationship of the semiconductor image; wherein, the target mask includes a semiconductor microstructure.

[0123] Optionally, the general-purpose semiconductor microstructure size measurement module is also specifically used for:

[0124] Semiconductor images acquired using an electron microscope are converted into semiconductor images in a standard computer format; the standard computer format semiconductor images include a scale bar.

[0125] The text portion of the scale bar and the bar diagram portion representing length are identified and detected to determine the pixel distance conversion relationship of the semiconductor image; wherein, the pixel distance conversion relationship is the ratio of the value of the text portion of the scale bar to the pixel value of the bar diagram portion.

[0126] Optionally, the general-purpose semiconductor microstructure size measurement module is also specifically used for:

[0127] The image segmentation result includes at least two masks, and overlapping mask removal processing is performed to obtain a first mask set; wherein, the overlapping mask removal processing includes obtaining overlapping masks, determining the area size of overlapping masks, and removing overlapping masks with smaller areas;

[0128] Noise in the first mask set is removed using image connected component analysis to obtain the second mask set; each mask in the second mask set includes only one semiconductor microstructure.

[0129] The noise masks in the second mask set are removed to obtain the third mask set.

[0130] The masks located at the edges of the semiconductor image in the third mask set are removed to obtain the fourth mask set;

[0131] Remove the masks located in the scale region from the fourth mask set to obtain the fifth mask set;

[0132] Remove the masks belonging to the background region from the fifth mask set to obtain the target mask set.

[0133] Optionally, the general-purpose semiconductor microstructure size measurement module is also specifically used for:

[0134] The pixel grayscale values ​​of the semiconductor image are determined, and a first threshold and a second threshold are determined based on the pixel grayscale values; wherein, the first threshold is greater than the second threshold, the first threshold is used to distinguish the brighter areas in the semiconductor image, and the second threshold is used to distinguish the darker areas in the semiconductor image.

[0135] The first mean and median of the image grayscale values ​​corresponding to the fifth mask are determined. If the median of the image grayscale values ​​of the fifth mask is greater than a first threshold, the fifth mask is determined to be a bright mask. If the median of the image grayscale values ​​of the fifth mask is less than a second threshold, the fifth mask is determined to be a dark mask. If the median of the image grayscale values ​​of the fifth mask is between the first and second thresholds, the second mean of the image grayscale values ​​within a preset range outside the fifth mask is determined. If the first mean is greater than the second mean, the fifth mask is a bright mask; otherwise, it is a dark mask. The set of fifth masks includes at least two fifth masks.

[0136] Determine the number of bright masks and dark masks in the fifth mask set. If the number of bright masks is greater than the number of dark masks, then the bright masks are used as the target mask set; if the number of bright masks is less than the number of dark masks, then the dark masks are used as the target mask set.

[0137] Optionally, the general-purpose semiconductor microstructure size measurement module is also specifically used for:

[0138] Determine the first and second moments of the target mask, and determine the principal and secondary axes of the target mask based on the first and second moments;

[0139] Determine the first pixel distance between the intersection point of the target mask's main axis and the target mask's contour, and determine the actual physical major axis of the target mask based on the first pixel distance and the pixel distance conversion relationship;

[0140] Determine the second pixel distance between the intersection point of the target mask sub-axis and the target mask contour, and determine the actual physical minor axis of the target mask based on the second pixel distance and the pixel distance conversion relationship.

[0141] Optionally, the general-purpose semiconductor microstructure size measurement module is also specifically used for:

[0142] If there are at least two sixth masks in the target mask that are obtained from the same position but different angles, then the minimum bounding rectangle of the at least two sixth masks is determined; the different angles include the shooting angle and the electron diffraction angle.

[0143] If the first cross-union ratio of the minimum bounding rectangles of the at least two sixth masks is greater than the first cross-union ratio threshold, then the second cross-union ratio of the at least two sixth masks is determined; if the second cross-union ratio is greater than the second cross-union ratio threshold, then the sixth mask with higher confidence is retained.

[0144] If the first intersection-union ratio of the minimum bounding rectangle of the at least two sixth masks is less than the first intersection-union ratio threshold, then the at least two sixth masks are retained simultaneously.

[0145] Optionally, the device further includes a correction module, specifically used for:

[0146] A display interface for microstructure size measurement results is constructed, and microstructure size measurement results with errors are corrected.

[0147] Optional, a dedicated semiconductor microstructure size measurement module, specifically used for:

[0148] The microstructure size measurement results are used as the training dataset;

[0149] The training dataset is input into the closed-set image segmentation model for model training, resulting in a converged target closed-set image segmentation model.

[0150] A dedicated semiconductor microstructure size measurement method is used to measure the size of the semiconductor to be measured in the target closed set image segmentation model;

[0151] Correspondingly, the method of using a dedicated semiconductor microstructure size measurement technique to measure the size of the semiconductor to be measured within the target closed-set image segmentation model includes:

[0152] The acquired semiconductor image to be measured is converted into an image format, and the pixel distance conversion relationship of the semiconductor image to be measured is determined.

[0153] A target closed-set image segmentation model is used to segment the image of the semiconductor to be measured and determine the image segmentation result; wherein, the image segmentation result includes at least two target masks, and each target mask includes a microstructure of the semiconductor to be measured;

[0154] Based on the pixel distance conversion relationship of the semiconductor image to be measured, the size measurement result of the target mask is determined.

[0155] The semiconductor microstructure size measurement device provided in the embodiments of the present invention can perform the semiconductor microstructure size measurement method provided in any of the embodiments of the present invention, and has the corresponding functions and beneficial effects of performing the semiconductor microstructure size measurement method. For details, please refer to the relevant operations of the semiconductor microstructure size measurement method in the foregoing embodiments.

[0156] Example 5

[0157] Figure 10 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0158] like Figure 10 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0159] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0160] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as semiconductor microstructure size measurement methods.

[0161] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0162] In some embodiments, the semiconductor microstructure size measurement method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the semiconductor microstructure size measurement method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the semiconductor microstructure size measurement method by any other suitable means (e.g., by means of firmware).

[0163] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transferring data and instructions to the storage system, the at least one input device, and the at least one output device.

[0164] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0165] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0166] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0167] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0168] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0169] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0170] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for measuring the dimensions of semiconductor microstructures, characterized in that, The method includes: A general semiconductor microstructure size measurement method is used to measure the size of the acquired semiconductor image to obtain the microstructure size measurement results. Among them, the general semiconductor microstructure size measurement method uses an open set image segmentation model to segment the semiconductor image. The open set image segmentation model supports segmentation of arbitrary semiconductor microstructures. The microstructure size measurement results are used as a training dataset and input into a closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method; wherein, the closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset; The step of measuring the size of the acquired semiconductor image using a general semiconductor microstructure size measurement method to obtain the microstructure size measurement result includes: The acquired semiconductor image is converted to an image format, and the pixel distance conversion relationship of the semiconductor image is determined. An open-set image segmentation model is used to segment a semiconductor image and determine the image segmentation result; wherein the image segmentation result includes at least two masks, and each mask includes at least one semiconductor microstructure; The image segmentation result includes at least two masks, which are preprocessed to obtain a target mask set; wherein the preprocessing includes removing overlapping masks, removing noise from the masks, removing noise masks, removing masks located at the image edges, removing masks in the image scale area, and removing background masks. The size measurement result of the target mask is determined based on the pixel distance conversion relationship of the semiconductor image; wherein, the target mask includes a semiconductor microstructure.

2. The method according to claim 1, characterized in that, The step of converting the acquired semiconductor image into an image format and determining the pixel distance conversion relationship of the semiconductor image includes: Semiconductor images acquired using an electron microscope are converted into semiconductor images in a standard computer format; the standard computer format semiconductor images include a scale bar. The text portion of the scale bar and the bar diagram portion representing length are identified and detected to determine the pixel distance conversion relationship of the semiconductor image; wherein, the pixel distance conversion relationship is the ratio of the value of the text portion of the scale bar to the pixel value of the bar diagram portion.

3. The method according to claim 1, characterized in that, The image segmentation result includes at least two masks, which are preprocessed to obtain a target mask set, including: The image segmentation result includes at least two masks, and overlapping mask removal processing is performed to obtain a first mask set; wherein, the overlapping mask removal processing includes obtaining overlapping masks, determining the area size of overlapping masks, and removing overlapping masks with smaller areas; Noise in the first mask set is removed using image connected component analysis to obtain the second mask set; each mask in the second mask set includes only one semiconductor microstructure. The noise masks in the second mask set are removed to obtain the third mask set. The masks located at the edges of the semiconductor image in the third mask set are removed to obtain the fourth mask set; Remove the masks located in the scale region from the fourth mask set to obtain the fifth mask set; Remove the masks belonging to the background region from the fifth mask set to obtain the target mask set.

4. The method according to claim 3, characterized in that, The step of removing the masks belonging to the background region from the fifth mask set to obtain the target mask set includes: The pixel grayscale values ​​of the semiconductor image are determined, and a first threshold and a second threshold are determined based on the pixel grayscale values; wherein, the first threshold is greater than the second threshold, the first threshold is used to distinguish the brighter areas in the semiconductor image, and the second threshold is used to distinguish the darker areas in the semiconductor image. The first mean and median of the image grayscale values ​​corresponding to the fifth mask are determined. If the median of the image grayscale values ​​of the fifth mask is greater than a first threshold, the fifth mask is determined to be a bright mask. If the median of the image grayscale values ​​of the fifth mask is less than a second threshold, the fifth mask is determined to be a dark mask. If the median of the image grayscale values ​​of the fifth mask is between the first and second thresholds, the second mean of the image grayscale values ​​within a preset range outside the fifth mask is determined. If the first mean is greater than the second mean, the fifth mask is a bright mask; otherwise, it is a dark mask. The set of fifth masks includes at least two fifth masks. Determine the number of bright masks and dark masks in the fifth mask set. If the number of bright masks is greater than the number of dark masks, then the bright masks are used as the target mask set; if the number of bright masks is less than the number of dark masks, then the dark masks are used as the target mask set.

5. The method according to claim 1, characterized in that, The determination of the target mask size measurement result based on the pixel distance conversion relationship of the semiconductor image includes: Determine the first and second moments of the target mask, and determine the principal and secondary axes of the target mask based on the first and second moments; Determine the first pixel distance between the intersection point of the target mask's main axis and the target mask's contour, and determine the actual physical major axis of the target mask based on the first pixel distance and the pixel distance conversion relationship; Determine the second pixel distance between the intersection point of the target mask sub-axis and the target mask contour, and determine the actual physical minor axis of the target mask based on the second pixel distance and the pixel distance conversion relationship.

6. The method according to claim 1, characterized in that, Before determining the size measurement result of the target mask based on the pixel distance conversion relationship of the semiconductor image, the method further includes: If there are at least two sixth masks in the target mask that are obtained from the same position but different angles, then the minimum bounding rectangle of the at least two sixth masks is determined; the different angles include the shooting angle and the electron diffraction angle. If the first cross-union ratio of the minimum bounding rectangles of the at least two sixth masks is greater than the first cross-union ratio threshold, then the second cross-union ratio of the at least two sixth masks is determined; if the second cross-union ratio is greater than the second cross-union ratio threshold, then the sixth mask with higher confidence is retained. If the first intersection-union ratio of the minimum bounding rectangle of the at least two sixth masks is less than the first intersection-union ratio threshold, then the at least two sixth masks are retained simultaneously.

7. The method according to claim 1, characterized in that, After obtaining the microstructure size measurement results by measuring the size of the acquired semiconductor image using a general semiconductor microstructure size measurement method within an open-set image segmentation model, the method further includes: A display interface for microstructure size measurement results is constructed, and microstructure size measurement results with errors are corrected.

8. The method according to claim 1, characterized in that, The step of inputting the microstructure size measurement results as a training dataset into a closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method, includes: The microstructure size measurement results are used as the training dataset; The training dataset is input into the closed-set image segmentation model for model training, resulting in a converged target closed-set image segmentation model. A dedicated semiconductor microstructure size measurement method is used to measure the size of the semiconductor to be measured in the target closed set image segmentation model; Correspondingly, the method of using a dedicated semiconductor microstructure size measurement technique to measure the size of the semiconductor to be measured within the target closed-set image segmentation model includes: The acquired semiconductor image to be measured is converted into an image format, and the pixel distance conversion relationship of the semiconductor image to be measured is determined. A target closed-set image segmentation model is used to segment the image of the semiconductor to be measured and determine the image segmentation result; wherein, the image segmentation result includes at least two target masks, and each target mask includes a microstructure of the semiconductor to be measured; Based on the pixel distance conversion relationship of the semiconductor image to be measured, the size measurement result of the target mask is determined.

9. A semiconductor microstructure dimension measuring device, characterized in that, The device includes: A general semiconductor microstructure size measurement module is used to measure the size of acquired semiconductor images using a general semiconductor microstructure size measurement method, and obtain the microstructure size measurement results. The general semiconductor microstructure size measurement method uses an open set image segmentation model to segment the semiconductor image, and the open set image segmentation model supports segmentation of arbitrary semiconductor microstructures. A dedicated semiconductor microstructure size measurement module is used to input the microstructure size measurement results as a training dataset into a closed-set image segmentation model for model training, so as to measure the size of the semiconductor microstructure to be measured using a dedicated semiconductor microstructure size measurement method; wherein, the closed-set image segmentation model is used to segment the semiconductor microstructures included in the training dataset; Specifically, the general-purpose semiconductor microstructure size measurement module is used for: The acquired semiconductor image is converted to an image format, and the pixel distance conversion relationship of the semiconductor image is determined. An open-set image segmentation model is used to segment a semiconductor image and determine the image segmentation result; wherein the image segmentation result includes at least two masks, and each mask includes at least one semiconductor microstructure; The image segmentation result includes at least two masks, which are preprocessed to obtain a target mask set; wherein the preprocessing includes removing overlapping masks, removing noise from the masks, removing noise masks, removing masks located at the image edges, removing masks in the image scale area, and removing background masks. The size measurement result of the target mask is determined based on the pixel distance conversion relationship of the semiconductor image; wherein, the target mask includes a semiconductor microstructure.

Images