Contour extraction methods and related products for SEM images

Abstract

This invention provides a method for contour extraction from SEM images and related products. The method includes acquiring an SEM image of the contour to be extracted and a corresponding design graphic; acquiring the line trend characteristics of the design graphic; placing multiple initial edge points in the SEM image based on the line trend characteristics; performing edge detection within a preset window centered on each initial edge point to obtain target edge points; and obtaining the target contour of the SEM image based on the multiple target edge points. This method adaptively allocates contour extraction data points from the SEM image, optimizes computational resources, improves the accuracy of contour extraction in geometrically complex regions, and avoids unnecessary consumption in geometrically simple regions, thereby improving the overall contour extraction efficiency.

Description

Technical Field

[0001] This invention relates to the field of integrated circuit technology, and in particular to a method for contour extraction of SEM images, a computer-readable storage medium, a computer program product, and a computer device. Background Technology

[0002] In integrated circuit manufacturing, high-precision contour extraction from scanning electron microscope (SEM) images is a crucial fundamental task. As semiconductor process nodes advance towards 7nm, 5nm, and even more advanced dimensions, the size of circuit patterns is approaching physical limits. Even minute deviations in shape and edges can directly lead to chip malfunction or significant performance degradation. Therefore, accurately extracting the contours of various patterns from SEM images characterizing wafer surface morphology is not only a necessary step in process monitoring but also a prerequisite for subsequent complex calculations and modeling.

[0003] In related technologies, contour extraction schemes for SEM images typically involve uniformly placing multiple edge-finding points at corresponding locations in the SEM image, and then using edge detection operators to search near these edge-finding points to find the locations of gray-level abrupt changes representing physical edges, which are then used as the final edge points to obtain the final contour. This contour extraction scheme is highly dependent on the location of the edge-finding points. A low density of edge-finding points may lead to insufficient contour extraction accuracy in complex graphics, while a high density consumes excessive computational resources and results in low contour extraction efficiency. Summary of the Invention

[0004] One object of the present invention is to provide a contour extraction method for SEM images, a computer-readable storage medium, a computer program product, and a computer device, so as to dynamically set edge finding points in SEM images, thereby improving the accuracy of contour extraction in complex graphics and improving the overall contour extraction efficiency.

[0005] A further objective of this invention is to enable dynamic setting of the range of a preset window for edge-finding points in SEM images, thereby improving the robustness and accuracy of contour extraction.

[0006] Specifically, according to one aspect of the present invention, the present invention provides a contour extraction method for SEM images, comprising: Obtain the SEM image of the contour to be extracted and the design graphic corresponding to the SEM image; Obtain the line trend characteristics of the designed graphic; Based on the line trend characteristics, multiple initial edge points are placed in the SEM image; Edge detection is performed within a preset window centered on each of the initial edge points to obtain the target edge points; The target contour of the SEM image is obtained based on multiple target edge points.

[0007] Optionally, the step of placing multiple initial edge points in the SEM image based on the line trend characteristics includes: Based on the characteristics of the line trend, multiple sampling points are placed on the design graphic; Each of the sampling points is mapped onto the SEM image to serve as the initial edge points.

[0008] Optionally, the line trend feature includes local curvature; and The step of placing multiple sampling points on the design graphic based on the line trend characteristics includes: Based on the local curvature of various parts of the design graphic, multiple sampling points are placed on the design graphic, wherein the placement density of the sampling points is positively correlated with the magnitude of the local curvature.

[0009] Optionally, the step of placing multiple sampling points on the design graphic based on the local curvature at various locations of the design graphic includes: Based on the local curvature of various parts of the design graphic, a plurality of segmentation points are determined on the lines of the design graphic at intervals, which are used as sampling points. The distance between two adjacent sampling points on the same line of the design graphic is negatively correlated with the magnitude of the local curvature.

[0010] Optionally, the step of performing edge detection within a preset window centered on the initial edge point to obtain the target edge point includes: Obtain the normal direction on the design graphic with the sampling point corresponding to the initial edge point as the tangent point, and map it into the SEM image as the edge search direction of the initial edge point; The edge search window width of the initial edge point is determined based on the line trend characteristics at the sampling point corresponding to the initial edge point; The preset window for the initial edge point is obtained based on the edge search direction and the edge search window width; Using an edge detection algorithm, the target edge point is determined within the preset window, starting from the initial edge point.

[0011] Optionally, determining the edge search window width of the initial edge point based on the line trend characteristics at the sampling point corresponding to the initial edge point includes: The edge search window width of the initial edge point is determined based on the local curvature at the sampling point corresponding to the initial edge point, wherein the edge search window width is positively correlated with the magnitude of the local curvature at the sampling point corresponding to the initial edge point.

[0012] Optionally, obtaining the target contour of the SEM image based on multiple target edge points includes: The target outline is formed by connecting multiple target edge points in sequence.

[0013] According to another aspect of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the steps of the contour extraction method for SEM images described above.

[0014] According to another aspect of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the steps of the contour extraction method for SEM images described above.

[0015] According to another aspect of the present invention, a computer device is also provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the contour extraction method for SEM images described above.

[0016] The contour extraction method for SEM images of the present invention obtains the complexity distribution of geometric shapes in various locations within the SEM image by acquiring the line trend characteristics of the design graphic. By placing initial edge points in the SEM image according to the line trend characteristics, the method achieves adaptive allocation of contour extraction data points in the SEM image, optimizes computational resources, improves the accuracy of contour extraction in geometrically complex areas, avoids unnecessary consumption in geometrically simple areas, and thus improves the overall contour extraction efficiency.

[0017] Furthermore, the contour extraction method for SEM images of the present invention adaptively sets the edge search window width of each initial edge point according to the line trend characteristics. On the one hand, it narrows the search range of simple geometric regions, accelerates the search process, and effectively avoids the situation of accidentally capturing the edges of adjacent lines. On the other hand, it expands the search range of complex geometric regions, effectively avoiding the situation of losing edges, thereby improving the robustness and accuracy of contour extraction as a whole.

[0018] The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments of the invention in conjunction with the accompanying drawings. Attached Figure Description

[0019] The following sections will describe some specific embodiments of the invention in detail by way of example and not limitation, with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar parts or portions. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 This is a schematic flowchart of a contour extraction method for SEM images according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the process of obtaining multiple initial edge points in a contour extraction method according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the edge search direction and edge search window width of a contour extraction method according to an embodiment of the present invention; Figure 4 This is a schematic flowchart illustrating edge detection within a preset window in a contour extraction method according to an embodiment of the present invention. Figure 5 This is a schematic diagram of a computer program product according to an embodiment of the present invention; Figure 6 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present invention; and Figure 7 This is a schematic diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0020] In integrated circuit manufacturing, high-precision contour extraction from scanning electron microscope (SEM) images is a crucial fundamental task. As semiconductor process nodes continue to advance to 7 nanometers, 5 nanometers, and even more advanced dimensions, the size of circuit patterns is approaching physical limits. Even minute deviations in their shape and edges can directly lead to chip malfunctions or significant performance degradation. Therefore, accurately extracting the contours of various patterns from SEM images characterizing wafer surface morphology is not only a necessary step in process monitoring but also a prerequisite for subsequent complex calculations and modeling.

[0021] Contour extraction from SEM images is not simply image binarization or edge detection. It requires algorithms to locate graphic boundaries at the sub-pixel level, accurately reflecting the complex distortions introduced by manufacturing processes such as optical proximity effects and etching load effects. Examples include corner rounding and local variations in critical dimensions (CD). This high-fidelity contour data forms a bridge connecting actual manufacturing results with chip design intent, and its accuracy directly determines the effectiveness and reliability of subsequent optimization and correction models.

[0022] Specifically, high-precision contours extracted from SEM images have a wide range of applications. Firstly, the core application of high-precision contours lies in the establishment and calibration of optical proximity correction (OPC) models. OPC technology is an indispensable part of current deep submicron integrated circuit manufacturing, aiming to counteract pattern distortion caused by diffraction, interference, and other effects during lithography by pre-correcting the mask pattern for inverse distortion. Building an accurate OPC model requires a large amount of paired data of "design pattern - actual silicon wafer image pattern" as a training basis. Here, the "actual silicon wafer image pattern" is obtained from SEM images through high-precision contour extraction. By comparing and analyzing these extracted contours with the original design data, the specific manifestations and intensities of various optical proximity effects can be quantitatively characterized, thereby training a mathematical model capable of predicting process deformation. Only based on high-precision input contours can the OPC model accurately simulate the imaging behavior of the lithography machine and generate effective correction schemes, ensuring that the final pattern obtained on the silicon wafer approximates the original design intent as closely as possible.

[0023] Furthermore, SEM image contours are also used to build other important models, such as etching process models and chemical mechanical polishing models. The etching process may further change the shape and sidewall angles of the pattern, and the contours extracted from the post-etching SEM image can be used to build and validate etching process simulation models and predict the differences in etching rates under different pattern densities and layouts.

[0024] In summary, high-precision profiles are the core data source driving the entire manufacturability design process and a key information carrier ensuring high yields of chips from design to finished product.

[0025] The purpose of the contour extraction method for SEM images in this embodiment is to dynamically set edge finding points in SEM images, thereby improving the accuracy of contour extraction in complex graphics and improving the overall contour extraction efficiency.

[0026] Figure 1 This is a flowchart illustrating a contour extraction method for SEM images according to an embodiment of the present invention. The method generally includes: S100, Obtain the SEM image of the contour to be extracted and the design graphic corresponding to the SEM image; S200, to obtain the line trend characteristics of the design graphic; S300: Based on the characteristics of line trends, multiple initial edge points are placed in the SEM image; S400 performs edge detection within a preset window centered on each initial edge point to obtain the target edge point; S500 obtains the target contour of the SEM image based on multiple target edge points.

[0027] SEM images can be photoresist images acquired after exposure and development in the photolithography process, or wafer etching images acquired after etching. SEM images are typically grayscale images. Design graphics can be feature graphics in a mask design layout, such as Manhattan polygons and Curvilinear photomasks. Manhattan polygons are polygons in a mask design layout consisting only of vertices, horizontal and vertical line segments, used to represent metal lines and other components in the chip. Curvilinear photomasks can contain curved shapes. Design graphics are typically binary images, stored in formats such as GDSII and OASIS.

[0028] In this embodiment, after obtaining the SEM image with extracted contours and its design graphic, the SEM image and the design graphic can be initially aligned (e.g., aligned based on coordinates) so that each region in the SEM image roughly maps to the design graphic one by one.

[0029] Next, a geometric analysis is performed on the lines of the design graphic to obtain the line trend characteristics at various locations, thereby determining the geometric change trend and complexity distribution of the design graphic. Specifically, when the design graphic includes a Manhattan pattern, the line trend characteristics may include straight line segments, corner segments, etc., where the geometry of straight line segments is relatively simple, while the geometry of corner segments (such as the corners of wires, line ends, or edges of densely arranged contact holes) is more complex. When the design graphic includes a curved mask pattern, the line trend characteristics may include local curvature, where the geometry is simpler where the local curvature is smaller, and more complex where the local curvature is larger.

[0030] Since each region in the SEM image roughly maps to a design graphic, mapping the line trend characteristics of the design graphic to the SEM image allows for a preliminary prediction of the approximate geometric change trend of the corresponding contours on the SEM image: which regions are straight or gently sloping line segments, and which regions exhibit drastic curvature changes. This step provides curvature analysis results for subsequent adaptive processing, enabling the algorithm to anticipate the complexity distribution throughout the SEM image.

[0031] Next, multiple initial edge points, or edge-finding points, are placed in the SEM image. In related techniques, initial edge points are typically placed evenly based on the outline lines of the design layout. The drawback of this approach is that in geometrically complex areas, such as graphic endpoints and complex 2D shapes, the curvature of the image outline changes significantly, while in geometrically simpler areas, such as straight lines, the change in image outline is minimal. If the density of initial edge points is the same in both geometrically complex and geometrically simple areas, the contour extraction accuracy in geometrically complex areas will be insufficient, while the extraction accuracy in geometrically simple areas will be too high. This results in inconsistent contour extraction accuracy across the entire image.

[0032] In this embodiment, multiple initial edge points are dynamically and adaptively placed in the SEM image based on the characteristics of the line trend. Specifically, in geometrically simple regions such as straight sections, areas with small curvature, or near-straight lines, relatively sparse initial edge points can be placed. In geometrically simple regions, fewer data points are needed to accurately describe the contour; oversampling only leads to redundant computation and reduces overall efficiency. By placing relatively sparse initial edge points in geometrically simple regions, computational resources can be saved and overall contour extraction efficiency improved while maintaining contour extraction accuracy. In geometrically complex regions, relatively dense initial edge points can be placed. Geometrically complex regions have rapidly changing contours and typically require more data points to reconstruct their true shape with high fidelity, avoiding the loss of crucial deformation details (such as corner radii and local depressions) due to insufficient sampling. By placing relatively dense initial edge points, the contour extraction accuracy in geometrically complex regions can be improved.

[0033] After placing the initial edge points, edge detection can be performed within a preset window centered on each initial edge point to obtain the target edge points. Specifically, edge detection operators (such as the Sobel operator, Canny operator, etc.) can be used to search within the preset window, starting from the initial edge points, to locate the point with the largest absolute gradient value, which is then taken as the target edge point. After determining the precise location of the target edge point within the preset window, the coordinate information of that point is recorded. After all target edge points have been found within the preset windows of all initial edge points, a series of discrete but high-precision two-dimensional point sets are formed to constitute the target contour.

[0034] Next, the target contour of the SEM image is reconstructed based on the obtained target edge points. Specifically, all target edge points corresponding to the same feature in the design drawing can be connected to form the target contour. Alternatively, the two-dimensional point set can be processed by interpolation, optimization, etc., before connecting the target edge points to form the target contour. After obtaining the target contour, it can be output in formats such as GDSII or OASIS (storing the target contour as a boundary layer), or in a text format specifically designed for measurement and modeling (such as a CSV or DAT file containing a series of X and Y coordinate pairs).

[0035] The contour extraction method for SEM images in this embodiment obtains the complexity distribution of geometric shapes in various locations within the SEM image by acquiring the line trend characteristics of the design graphic. By dynamically and adaptively placing initial edge points in the SEM image based on the line trend characteristics, the method achieves on-demand allocation of contour extraction data points, optimizes computational resources, improves the accuracy of contour extraction in geometrically complex areas, and avoids unnecessary consumption in geometrically simple areas, thereby improving the overall contour extraction efficiency.

[0036] In some embodiments of the contour extraction method for SEM images of the present invention, such as Figure 2 As shown, based on the characteristics of the line trend, multiple initial edge points are placed in the SEM image, including: S311, based on the characteristics of the line trend, places multiple sampling points on the design graphic; S313 maps each sampling point onto the SEM image as initial edge points.

[0037] In this embodiment, multiple sampling points are first placed on the design graphic and then mapped to the SEM image, thereby improving the efficiency and accuracy of the initial edge point placement.

[0038] For example, multiple sampling points can be directly set on the lines of the design graphic based on the line trend characteristics at various locations within the design graphic. Specifically, when the design graphic includes a Manhattan shape, sparser sampling points are placed in the straight sections of the Manhattan shape, while denser sampling points are placed in the corner sections. When the design graphic includes a curved mask shape, sparser sampling points are placed in areas with lower local curvature, while denser sampling points are placed in areas with higher local curvature.

[0039] For example, after adjusting the design graphic (e.g., rounding corners, or extending specific lines inward or outward by a preset distance), multiple sampling points can be set on the adjusted lines based on their trend characteristics. This can further improve the placement accuracy of the sampling points.

[0040] In some embodiments of the contour extraction method for SEM images of the present invention, the line trend features include local curvature; based on the line trend features, multiple sampling points are placed on the design graphic, including: Based on the local curvature of various parts of the design graphic, multiple sampling points are placed on the design graphic, and the placement density of the sampling points is positively correlated with the magnitude of the local curvature.

[0041] In this embodiment, the local curvature can be the curvature at each point of the line in the curve mask graphic, or it can be the curvature at each point of the line after rounding the corners in the Manhattan graphic.

[0042] Sampling points can be placed directly on the lines of the design graphic, or they can be placed on the adjusted lines.

[0043] On the closed lines of the same feature pattern, the placement density of sampling points is greater where the local curvature is larger than that where the local curvature is smaller. For example, the placement density of sampling points can be proportional to their local curvature.

[0044] In some embodiments of the contour extraction method for SEM images of the present invention, multiple sampling points are placed on the design graphic based on the local curvature of various locations therein, including: Based on the local curvature of various parts of the design graphic, multiple segmentation points are determined on the lines of the design graphic at intervals, serving as sampling points. Among them, on the same line of the design graphic, the distance between two adjacent sampling points is negatively correlated with the magnitude of the local curvature.

[0045] In this embodiment, sampling points are directly set on the lines of the design graphic. The distance between two adjacent sampling points can represent the shortest distance between the two sampling points along the line they lie on. For example, the distance between two adjacent sampling points can be calculated using the following formula: D X =D a / (N*C X ), Among them, D X D represents the distance between two adjacent sampling points. a C represents the average distance between multiple sets of adjacent sampling points. X This represents the local curvature at the sampling point, and N is the spacing reduction factor, which can be set by the user as needed.

[0046] Please refer to Figure 3 , Figure 3 This illustrates the case where the design graphic and the SEM graphic after the contour extraction are aligned and stacked. Figure 3The middle line 61 represents a part of the lines of a feature graphic in the design graphic, the outline 62 represents the outline extracted from the SEM graphic, the shading inside the outline 62 represents the groove shape in the SEM graphic, and points 41 and 51 represent the sampling points of the design graphic (which are mapped to the initial edge points in the SEM graphic). Figure 3 In the sample, sampling point 41 is located in a region with relatively small local curvature, while sampling point 51 is located in a region with relatively large local curvature. The spacing between sampling points 41 is greater than the spacing between sampling points 51.

[0047] In some embodiments of the contour extraction method for SEM images of the present invention, such as Figure 4 As shown, edge detection is performed within a preset window centered on the initial edge point to obtain the target edge points, including: S411: Obtain the normal direction of the initial edge point with the sampling point corresponding to the initial edge point as the tangent point on the design graphic, and map it into the SEM image as the edge search direction of the initial edge point; S413, determine the edge search window width of the initial edge point based on the line trend characteristics at the sampling points corresponding to the initial edge point; S415, Based on the edge search direction and the edge search window width, obtain the preset window for the initial edge points; S417 uses an edge detection algorithm to determine the target edge point within a preset window, starting from the initial edge point.

[0048] In addition to accurately placing each initial edge point, it is also necessary to accurately set the preset window for each initial edge point to achieve high-precision contour extraction. Whether the preset window setting is reasonable can directly determine the success or failure of contour extraction.

[0049] The default window is usually a one-dimensional search path window, which includes the edge search direction and the edge search window width. The edge search direction is used to determine the search path direction, and the edge search window width is used to determine the search path length.

[0050] The edge search direction is typically the normal direction with the sampling point corresponding to the initial edge point as the tangent point. For straight sections, the edge search direction is its perpendicular direction. For curved sections, the edge search direction is the normal direction of the sampling point.

[0051] Setting the edge search window width too long or too short can introduce serious errors. If the edge search window is too long, when the algorithm searches for extreme points of gray-level gradient changes (i.e., target edge points) along the edge search direction, its search range may exceed the true edge of the target graphic, thus capturing the edge of adjacent graphics (such as another parallel line), or misinterpreting gray-level changes caused by image noise or material contrast differences as graphic boundaries. This "over-search" phenomenon is particularly prominent in advanced process nodes with densely arranged graphics and extremely small spacing, which can easily lead to severe distortion of the extracted contours, or even incorrectly connect or merge two independent lines, completely distorting the topological structure of the graphic, causing subsequent modeling work to be based on an incorrect foundation.

[0052] Conversely, if the edge search window width is set too short, the algorithm may fail to cover the true edge locations. Due to noise in the SEM image itself, edge effects (such as bright edge bands), and the potential for gradual transitions in the image, the true edge gradient peaks may be located several pixels beyond the estimated location. An excessively short search range may cause the algorithm to terminate the search prematurely before finding a clear gradient extremum, thus missing edges. This can result in incomplete contour extraction, local missing features, or inward shrinkage, failing to accurately reflect the actual size and shape of the image, especially underestimating line widths or misjudging the curvature of rounded corners.

[0053] In this embodiment, the edge search window width for each initial edge point is dynamically and adaptively determined based on the line trend characteristics. Specifically, a narrower edge search window width can be used in geometrically simple regions such as straight sections, areas with small curvature, or near-straight lines. Within geometrically simple regions, the contour position is relatively stable and predictable. Using a narrower edge search window width not only accelerates the search process but, more importantly, effectively avoids mis-capturing the edges of adjacent lines in dense graphics due to an excessively wide search range, thus greatly improving the robustness and accuracy of contour extraction. In geometrically complex regions such as corners or ends with large curvature, a wider edge search window width can be used. Contours change rapidly within geometrically complex regions, and process variations (such as the proximity effect of lithography-etching) are often more significant. The actual imaged edge position may deviate significantly from the design expectation, and the edge gradient may be dispersed due to deformation. Using a wider edge search window width ensures that the algorithm has sufficient field of view to capture real edges that may drift significantly, preventing edge loss.

[0054] In some embodiments of the contour extraction method for SEM images of the present invention, the edge search window width of the initial edge point is determined based on the line trend characteristics at the sampling points corresponding to the initial edge point, including: The edge search window width of the initial edge point is determined based on the local curvature at the sampling point corresponding to the initial edge point. The edge search window width is positively correlated with the magnitude of the local curvature at the sampling point corresponding to the initial edge point.

[0055] For example, the width of the edge search window for the initial edge points can be calculated using the following formula: R X =R a *M*C X , Among them, R X R represents the width of the edge search window for the initial edge points. a C represents the average edge search window width across multiple initial edge points. X This represents the local curvature at the sampling point corresponding to the initial edge point, and M is the search range magnification factor, which can be set by the user as needed.

[0056] Please continue to refer to this. Figure 3 , Figure 3 The dashed line 42 with arrows indicates the preset window of sampling point 41, and the dashed line 52 with arrows indicates the preset window of sampling point 51. The arrows indicate the direction of edge search, and the length of the dashed line segment indicates the width of the edge search window. Figure 3 In the image, sampling point 41, corresponding to dashed line 42, is located in a region with low local curvature, while sampling point 51, corresponding to dashed line 52, is located in a region with high local curvature. The dashed line segment of dashed line 42 is shorter than that of dashed line 52. If sampling point 51 uses the same edge search window width as sampling point 41, the contour edges at rounded corners of the SEM image may be lost.

[0057] In some embodiments of the SEM image contour extraction method of the present invention, the target contour of the SEM image is obtained based on multiple target edge points, including: Connect multiple target edge points sequentially to form the target outline.

[0058] In this embodiment, based on the positional relationship of the sampling points corresponding to each target edge point, the target edge points corresponding to multiple sampling points in the same feature pattern can be connected sequentially to form a target contour corresponding to the feature pattern.

[0059] The flowchart provided in this embodiment is not intended to indicate that the operations of the method will be performed in any particular order, or that all operations of the method are included in every case. Furthermore, the method may include additional operations. Within the scope of the technical concept provided by the method in this embodiment, additional variations can be made to the above method.

[0060] It should be understood that in some embodiments, the components may be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods may be implemented using software or firmware stored in memory and executed by a suitable instruction execution system.

[0061] This invention also provides a computer program product 10, a computer-readable storage medium 20, and a computer device 30. Figure 4 This is a schematic diagram of a computer program product 10 according to an embodiment of the present invention. Figure 5 This is a schematic diagram of a computer-readable storage medium 20 according to an embodiment of the present invention. Figure 6 This is a schematic diagram of a computer device 30 according to an embodiment of the present invention. The computer program product 10 includes a computer program 11, which, when executed by the processor 32, implements the steps of the contour extraction method for SEM images described above. A computer-readable storage medium 20 stores the computer program 11 thereon, which, when executed by the processor 32, implements the steps of the contour extraction method for SEM images described above. The computer device 30 may include a memory 31, a processor 32, and the computer program 11 stored in the memory 31 and running on the processor 32.

[0062] The computer program 11 used to perform the operations of this invention may be assembly instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, integrated circuit configuration data, or source code or object code written in any combination of one or more programming languages ​​and procedural programming languages. The computer program 11 may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer may be connected to the user's computer via any type of network, including a Local Area Network (LAN) or Wide Area Network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, to perform aspects of this invention, electronic circuits, including, for example, programmable logic circuits, Field-Programmable Gate Arrays (FPGAs), or Programmable Logic Arrays (PLAs), may execute computer-readable program instructions to personalize the electronic circuits by utilizing state information from computer-readable program instructions.

[0063] For the purposes of this embodiment, computer program product 10 is a related product that includes computer program 11.

[0064] For the purposes of this embodiment, the computer-readable storage medium 20 is a tangible device capable of holding and storing a computer program 11. It can be any device capable of containing, storing, communicating, propagating, or transmitting the computer program 11 for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable storage medium 20 include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable optical disc read-only memory (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical encoding device, and any suitable combination thereof.

[0065] Computer device 30 can be, for example, a server, desktop computer, laptop computer, tablet computer, or smartphone. In some examples, computer device 30 can be a cloud computing node. Computer device 30 can be described in the general context of computer system executable instructions (such as program modules) executed by a computer system. Typically, program modules can include routines, programs, object programs, components, logic, data structures, etc., that perform specific tasks or implement specific abstract data types. Computer device 30 can be implemented in a distributed cloud computing environment where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can reside on local or remote computing system storage media, including storage devices.

[0066] Computer device 30 may include a processor 32 adapted to execute stored instructions and a memory 31 that provides temporary storage space for the operation of said instructions during operation. The processor 32 may be a single-core processor, a multi-core processor, a computing cluster, or any other configuration. The memory 31 may include random access memory (RAM), read-only memory, flash memory, or any other suitable storage system.

[0067] Computer device 30 may also include a network adapter / interface and an input / output (I / O) interface. The I / O interface allows external devices that can be connected to the computer device to input and output data. The network adapter / interface provides communication between the computer device and a network, typically represented as a communication network.

[0068] Therefore, those skilled in the art should recognize that although numerous exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications conforming to the principles of the present invention can be directly determined or derived from the disclosure of the present invention without departing from the spirit and scope of the invention. Thus, the scope of the present invention should be understood and construed as covering all such other variations or modifications.

Claims

1. A method for contour extraction from SEM images, characterized in that, include: Obtain the SEM image of the contour to be extracted and the design graphic corresponding to the SEM image; Obtain the line trend characteristics of the designed graphic; Based on the line trend characteristics, multiple initial edge points are placed in the SEM image; Edge detection is performed within a preset window centered on each of the initial edge points to obtain the target edge points; The target contour of the SEM image is obtained based on multiple target edge points.

2. The contour extraction method according to claim 1, characterized in that, The step of placing multiple initial edge points in the SEM image based on the line trend characteristics includes: Based on the characteristics of the line trend, multiple sampling points are placed on the design graphic; Each of the sampling points is mapped onto the SEM image to serve as the initial edge points.

3. The contour extraction method according to claim 2, characterized in that, The line trend characteristics include local curvature; and The step of placing multiple sampling points on the design graphic based on the line trend characteristics includes: Based on the local curvature of various parts of the design graphic, multiple sampling points are placed on the design graphic, wherein the placement density of the sampling points is positively correlated with the magnitude of the local curvature.

4. The contour extraction method according to claim 3, characterized in that, The step of placing multiple sampling points on the design graphic based on the local curvature at various locations includes: Based on the local curvature of various parts of the design graphic, a plurality of segmentation points are determined on the lines of the design graphic at intervals, which are used as sampling points. The distance between two adjacent sampling points on the same line of the design graphic is negatively correlated with the magnitude of the local curvature.

5. The contour extraction method according to claim 4, characterized in that, The step of performing edge detection within a preset window centered on the initial edge point to obtain the target edge point includes: Obtain the normal direction on the design graphic with the sampling point corresponding to the initial edge point as the tangent point, and map it into the SEM image as the edge search direction of the initial edge point; The edge search window width of the initial edge point is determined based on the line trend characteristics at the sampling point corresponding to the initial edge point; The preset window for the initial edge point is obtained based on the edge search direction and the edge search window width; Using an edge detection algorithm, the target edge point is determined within the preset window, starting from the initial edge point.

6. The contour extraction method according to claim 5, characterized in that, The step of determining the edge search window width of the initial edge point based on the line trend characteristics at the sampling point corresponding to the initial edge point includes: The edge search window width of the initial edge point is determined based on the local curvature at the sampling point corresponding to the initial edge point, wherein the edge search window width is positively correlated with the magnitude of the local curvature at the sampling point corresponding to the initial edge point.

7. The contour extraction method according to claim 1, characterized in that, The step of obtaining the target contour of the SEM image based on multiple target edge points includes: The target outline is formed by sequentially connecting multiple target edge points.

8. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the contour extraction method for SEM images as described in any one of claims 1 to 7.

9. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the contour extraction method for SEM images as described in any one of claims 1 to 7.

10. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the contour extraction method for SEM images according to any one of claims 1 to 7.

Images