Mask true image gray scale quantification method and related product
By using photoelectric response models and spatial registration techniques, the problem of grayscale range distortion between simulated and measured images was solved, achieving higher pixel value alignment accuracy and comparability between simulated and measured spatial images, thus improving detail matching accuracy and registration efficiency.
Patent Information
- Application Number
- CN202610099989.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-23
- Publication Date
- 2026-05-01
AI Technical Summary
In the existing technology, during the pixel value comparison process between the simulated image and the measured image, the grayscale range is improperly compressed or stretched, resulting in distortion and affecting the comparability of the simulated spatial image and the measured spatial image.
By using the photoelectric response model of the image acquisition device, the measured spatial image is restored into an actual light intensity image. The simulated spatial image is calibrated using the simulated light intensity values at multiple preset positions and the incident light intensity values at the corresponding measured positions. A corrected light intensity image is constructed and spatial registration is performed to eliminate grayscale errors and improve the accuracy of pixel value calibration.
This study solves the problem of grayscale distortion in simulated spatial images, improves the comparability and alignment accuracy between simulated and measured spatial images, eliminates grayscale errors introduced by pixel-level misalignment, and improves detail matching accuracy and spatial registration efficiency.
Smart Images
Figure CN121956441A_ABST
Abstract
Description
Gray quantization methods and related products for masking realistic images Technical Field
[0001] This invention relates to the field of semiconductor photolithography simulation technology, and in particular to a grayscale measurement method for masked simulation images and related products. Background Technology
[0002] In the field of semiconductor optical inspection, the comparative analysis of simulated and measured images (e.g., spatial images acquired through a charge-coupled device (CCD)) of a mask's spatial image is a crucial step in process verification and defect detection. Among these, pixel value calibration between the simulated and measured images is one of the key aspects ensuring the accuracy of the comparative analysis results. Currently, the Global Linear Normalization (GLN) method is commonly used for pixel value calibration, which directly utilizes the global maximum value i from the simulated image. M and the minimum value of the entire region i m Linear scaling is applied to the entire simulation image: Among them, i s,g (x, y) represents the gray value at point (x, y) in the simulated grayscale image corresponding to the simulated image, i s,r (x, y) represents the simulated light intensity value at point (x, y) in the simulated image, i M and i m These represent the maximum and minimum light intensity values in the simulated light intensity values of the simulated image, respectively. In the above scheme, the maximum and minimum light intensity values in the simulated light intensity values are converted into grayscale values of 255 and 0, respectively.
[0003] However, due to physical conditions and extreme values of the entire image (such as noise and background drift), using global linear mapping can cause the grayscale range of the simulated grayscale image to be improperly compressed or stretched, leading to distortion. Summary of the Invention
[0004] In view of the above problems, the present invention is proposed to provide a grayscale measurement method and related products for masked realistic images that overcome or at least partially solve the above problems.
[0005] One objective of this invention is to address the problem of distortion caused by improper compression or stretching of the grayscale range during pixel value comparison between simulated and measured spatial images of a mask, thereby improving the accuracy of pixel value comparison and enhancing the comparability between simulated and measured spatial images.
[0006] Specifically, according to one aspect of the present invention, the present invention provides a grayscale quantization method for a mask-simulated image, comprising: simulating a target mask to obtain a simulated spatial image; obtaining a measured spatial image of the target mask acquired by an image acquisition device; using a photoelectric response model to restore the measured pixel values of the measured spatial image to incident light intensity values to obtain an actual light intensity image, wherein the photoelectric response model is configured to characterize the photoelectric response characteristics of the image acquisition device; obtaining simulated light intensity values at multiple preset positions in the simulated spatial image, and obtaining the incident light intensity values at measured positions in the actual light intensity image corresponding to each of the preset positions; calibrating the simulated light intensity values at various locations in the simulated spatial image using the simulated light intensity values at the multiple preset positions and the incident light intensity values corresponding to the measured positions to obtain a corrected light intensity image; and spatially registering the corrected light intensity image and the actual light intensity image.
[0007] Optionally, the photoelectric response model construction steps include: acquiring multiple test images obtained by the image acquisition device under various exposure conditions, and obtaining multiple sets of experimental data on exposure intensity and corresponding pixel values of the test images; determining the response critical value of the image acquisition device by analyzing the multiple sets of experimental data; wherein, when the exposure intensity is less than the response critical value, the exposure intensity and the corresponding pixel value of the test image have a linear relationship; and when the exposure intensity is greater than or equal to the response critical value, the exposure intensity and the corresponding pixel value of the test image have a non-linear relationship; using the multiple sets of experimental data, with the response critical value as the interval boundary point, fitting a linear relationship interval function and a non-linear relationship interval function regarding the exposure intensity and the corresponding pixel value of the test image; and constructing the photoelectric response model based on the linear relationship interval function and the non-linear relationship interval function.
[0008] Optionally, the step of obtaining simulated light intensity values at multiple preset positions in the simulated spatial image and obtaining the incident light intensity values at measured positions corresponding to each preset position in the actual light intensity image includes: obtaining the central region of the high transmittance region of the target mask; determining a highlight region in the simulated spatial image corresponding to the central region of the high transmittance region; obtaining the simulated light intensity value of the highlight region and obtaining the incident light intensity value at the measured position corresponding to the highlight region in the actual light intensity image; obtaining the central region of the low transmittance region of the target mask; determining a low-light region in the simulated spatial image corresponding to the central region of the low transmittance region; obtaining the simulated light intensity value of the low-light region and obtaining the incident light intensity value at the measured position corresponding to the low-light region in the actual light intensity image.
[0009] Optionally, the step of calibrating the simulated light intensity values at various locations in the simulated spatial image using the simulated light intensity values at multiple preset locations and the incident light intensity values at the corresponding measured locations to obtain a corrected light intensity image includes: using the incident light intensity value at the measured location corresponding to the highlight region as the corrected light intensity value of the simulated light intensity value of the highlight region, using the incident light intensity value at the measured location corresponding to the low-light region as the corrected light intensity value of the simulated light intensity value of the low-light region, constructing a linear regression model of the corrected light intensity value with respect to the simulated light intensity value; and using the linear regression model to calibrate the simulated light intensity values at various locations in the simulated spatial image to correspond to the corrected light intensity values to obtain the corrected light intensity image.
[0010] Optionally, the spatial registration of the corrected light intensity image and the actual light intensity image includes: converting the corrected light intensity image into a pixelated image corresponding to the actual light intensity image; and performing sub-pixel-level spatial registration of the pixelated image and the actual light intensity image.
[0011] Optionally, converting the corrected light intensity image into a pixelated image corresponding to the actual light intensity image includes: dividing the corrected light intensity image into multiple local regions, each local region corresponding one-to-one with a pixel in the actual light intensity image; performing a regional integral calculation of the corrected light intensity value on each local region, and using the calculation result as the equivalent pixel value of the local region to obtain the pixelated image, wherein the corrected light intensity value is the simulated light intensity value after calibration.
[0012] Optionally, the subpixel-level spatial registration of the pixelated image and the actual light intensity image includes: performing subpixel-level displacement on the pixelated image according to multiple translation amounts to obtain a shifted image corresponding to each translation amount; wherein, for each equivalent pixel in each shifted image, the target light intensity value of each equivalent pixel is obtained by interpolating the corrected light intensity value of the neighborhood region corresponding to the pixel center of each equivalent pixel in the pixelated image; based on the correlation analysis of mutual information, the target light intensity value of each equivalent pixel in each shifted image that has the highest correlation with the incident light intensity value of each pixel in the actual light intensity image is selected as the target registration image.
[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 grayscale measurement method for any of the above-described mask-like images.
[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 grayscale measurement method for any of the above-described mask-like images.
[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 grayscale measurement method for any of the above-described mask-like images.
[0016] The grayscale measurement method for masked simulated images of the present invention restores the measured spatial image to an actual light intensity image by using the photoelectric response model of the image acquisition device, thus solving the problem of improper compression or stretching of the grayscale range of pixels in the measured spatial image. By using simulated light intensity values at multiple preset locations and the incident light intensity values at corresponding measured locations, the simulated spatial image is calibrated to obtain a corrected light intensity image, solving the problem of distortion of simulated light intensity values at various locations in the simulated spatial image. Spatial registration using the corrected light intensity image and the actual light intensity image improves the accuracy of grayscale measurement and enhances the comparability between the simulated spatial image and the measured spatial image.
[0017] Furthermore, the grayscale quantization method of the present invention significantly eliminates grayscale errors (such as edge blurring and peak shift) caused by pixel-level misalignment by performing sub-pixel-level affine transformation on the pixelated image, thereby improving detail matching accuracy and greatly enhancing spatial registration efficiency.
[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 description will focus on some specific embodiments of the invention 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 components. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: FIG1 is a flowchart illustrating a grayscale method for masking a simulated image according to an embodiment of the invention; FIG2 is a flowchart illustrating obtaining a photoelectric response model using a grayscale method according to an embodiment of the invention; FIG3 is a flowchart illustrating obtaining a benchmark using a grayscale method according to an embodiment of the invention; FIG4 is a flowchart illustrating benchmarking simulated light intensity values using a grayscale method according to an embodiment of the invention; FIG5 is a flowchart illustrating pixelation and spatial registration of a corrected light intensity image using a grayscale method according to an embodiment of the invention; FIG6 is a flowchart illustrating pixelation of a corrected light intensity image using a grayscale method according to an embodiment of the invention; FIG7 is a flowchart illustrating spatial registration of a pixelated image using a grayscale method according to an embodiment of the invention; FIG8 is a schematic diagram of a computer program product according to an embodiment of the invention; FIG9 is a schematic diagram of a computer-readable storage medium according to an embodiment of the invention; and FIG10 is a schematic diagram of a computer device according to an embodiment of the invention. Detailed Implementation
[0020] The purpose of the grayscale quantization method for the masked simulated image in this embodiment is to solve the problem that the grayscale range is improperly compressed or stretched during the pixel value comparison process between the simulated spatial image and the measured spatial image of the mask, which leads to distortion. This method aims to improve the accuracy of pixel value comparison and enhance the comparability between the simulated spatial image and the measured spatial image.
[0021] Figure 1 is a flowchart illustrating a grayscale quantization method for a mask-simulated image according to an embodiment of the present invention. The method generally includes: S100, simulating a target mask to obtain a simulated spatial image; S200, obtaining a measured spatial image of the target mask acquired by an image acquisition device; S300, using a photoelectric response model to restore the measured pixel values of the measured spatial image to incident light intensity values, obtaining an actual light intensity image, wherein the photoelectric response model is configured to characterize the photoelectric response characteristics of the image acquisition device; S400, obtaining simulated light intensity values at multiple preset positions in the simulated spatial image, and obtaining incident light intensity values at measured positions corresponding to each preset position in the actual light intensity image; S500, using the simulated light intensity values at multiple preset positions and the incident light intensity values at corresponding measured positions to calibrate the simulated light intensity values at various locations in the simulated spatial image, obtaining a corrected light intensity image; S600, spatially registering the corrected light intensity image and the actual light intensity image.
[0022] A target mask is a mask used for photolithography on wafers during integrated circuit manufacturing. An aerial image (AI) is the light intensity distribution of an image formed by a mask pattern on an ideal image plane of an optical lithography system.
[0023] In the process of optical simulation of a target mask, a simulation model (e.g., a physical optics model) is used to calculate the exposure image of the target mask pattern on an ideal image plane under preset optical parameters, thus obtaining a simulated spatial image. Then, under experimental conditions, the target mask is exposed using actual optical parameters identical to the preset parameters, and a measured spatial image is acquired using an image acquisition device (e.g., a charge-coupled device). By comparing the simulated and measured spatial images, reverse calculations are performed to improve the relevant parameters in the simulation model until the image results expressed by the simulated and measured spatial images are consistent.
[0024] Images acquired by charge-coupled devices (CCDs) (i.e., measured spatial images), also known as CCD images, are typically pixelated grayscale images, where the grayscale values of the measured pixels usually range from 0 to 255. Simulated spatial images obtained from simulation models are usually continuous distribution maps of relative electric field intensity or relative light intensity, where the simulated light intensity values range, for example, from 0 to 1. Therefore, when comparing simulated and measured spatial images, they need to be converted to a reference value of the same range first.
[0025] As described in the background section, existing technologies typically use global linear mapping (also known as direct extremum scaling) to linearly scale the simulated light intensity values at various points in the simulated spatial image to the range [0, 255]. Due to physical conditions and extreme values of the entire image (such as noise and background drift), using global linear mapping can cause the grayscale range of the simulated spatial image to be improperly compressed or stretched, leading to distortion.
[0026] In this embodiment, the incident light intensity value is the light intensity value of the light incident on the charge-coupled device (CCD), and the measured pixel value is the light intensity value output by the CCD after photoelectric conversion. The photoelectric response model between the measured pixel value and the incident light intensity value reflects the photoelectric response characteristics of the CCD, that is, the mapping relationship between the power of the incident beam and the output current. On the one hand, since the beam used for photolithography is usually a monochromatic and highly coherent laser, the influence of different frequencies on the conversion signal of the CCD is ignored. On the other hand, the output current can be directly converted into a pixel value. Therefore, the photoelectric response model of the CCD in this embodiment can be simplified to the mapping relationship between the light intensity of the incident beam and the output pixel value.
[0027] It should be noted that in charge-coupled devices (CCDs), the charge amount of a single pixel has a full-well capacity threshold. When the charge amount approaches the full-well capacity, the photoelectric response enters the nonlinear region. This is because the number of electrons exceeds the storage capacity of a single pixel, causing signal truncation or diffusion to adjacent pixels, forming a halo. Generally, in the nonlinear region, the slope of the light intensity-pixel value curve in the photoelectric response model gradually decreases. In other words, as the light intensity increases, the pixel value difference corresponding to the same light intensity difference will be compressed.
[0028] Unlike existing technologies that linearly scale the simulated light intensity values at various points in the simulated spatial image to the [0, 255] range, this embodiment does not limit the range of simulated light intensity values in the simulated spatial image. Instead, it uses a photoelectric response model to restore the measured pixel values at various points in the measured spatial image to the incident light intensity values, obtaining an actual light intensity image. On the one hand, the actual light intensity image better reflects the true incident light intensity values, which can improve the accuracy of subsequent spatial registration and comparison. On the other hand, it avoids the situation where the measured pixel values corresponding to the incident light intensity values in the nonlinear region of the measured spatial image are compressed, thereby avoiding the situation where the calculation weight of the measured pixel values in the nonlinear region is reduced due to numerical compression during subsequent spatial registration and comparison, thus improving the accuracy of subsequent spatial registration and comparison.
[0029] As mentioned earlier, the simulated light intensity values at various locations in the simulated spatial image are typically relative light intensities. To compare them with actual light intensity images, the simulated spatial image needs to be calibrated for light intensity values. In this embodiment, the simulated light intensity values at multiple preset locations in the simulated spatial image and the incident light intensity values at corresponding measured locations in the actual light intensity image are used as calibration benchmarks to calibrate the simulated spatial image and obtain a corrected light intensity image. The preset locations can be areas with minimal noise and background drift, thereby improving the accuracy of the calibration benchmark. Preferably, the preset locations may include positions near the upper limit and lower limit of the simulated light intensity values in the simulated spatial image, allowing the calibration benchmark to reflect the overall deviation and improving calibration accuracy.
[0030] Since the actual light intensity image has already corrected for the nonlinear defects of the charge-coupled device, a benchmark can be used to linearly scale the simulated space image, accurately correcting the simulated light intensity values at various points in the simulated space image to the corrected light intensity values. Benchmarking can eliminate or reduce light intensity deviations caused by optical parameter errors, improving the simulated space image and making the corrected light intensity image more accurately reflect the incident light intensity value.
[0031] Next, spatial registration is performed on the calibrated light intensity image and the actual light intensity image, so that the calibrated light intensity value at each point in the calibrated light intensity image corresponds one-to-one with the incident light intensity value of each pixel in the actual light intensity image, thereby enabling a comparison of the differences between the two.
[0032] This embodiment's method for grayscale quantization of masked simulated images restores the measured spatial image to an actual light intensity image using the photoelectric response model of the image acquisition device, thus solving the problem of improper compression or stretching of pixel grayscale ranges in the measured spatial image. This embodiment calibrates the simulated spatial image by using simulated light intensity values at multiple preset locations and the corresponding incident light intensity values at the measured locations, obtaining a corrected light intensity image, thus solving the problem of distortion of simulated light intensity values at various points in the simulated spatial image. This embodiment improves the accuracy of pixel value alignment and enhances the comparability between the simulated and measured spatial images by performing spatial registration using the corrected light intensity image and the actual light intensity image.
[0033] In some embodiments of the gray quantification method of the present invention, as shown in Figure 2, the steps for constructing the photoelectric response model include: S311, acquiring multiple test images obtained by an image acquisition device under various exposure conditions, and obtaining multiple sets of experimental data on exposure intensity and corresponding pixel values of the test images; S313, determining the response critical value of the image acquisition device by analyzing multiple sets of experimental data; wherein, when the exposure intensity is less than the response critical value, the exposure intensity and the pixel value of the corresponding test image have a linear relationship; and when the exposure intensity is greater than or equal to the response critical value, the exposure intensity and the pixel value of the corresponding test image have a non-linear relationship; S315, using multiple sets of experimental data, with the response critical value as the interval boundary point, fitting a linear relationship interval function and a non-linear relationship interval function regarding the exposure intensity and the pixel value of the corresponding test image; S317, constructing a photoelectric response model based on the linear relationship interval function and the non-linear relationship interval function.
[0034] This embodiment is used to obtain the photoelectric response model of an image acquisition device. Specifically, a stepped light intensity experiment can be conducted, gradually increasing the exposure intensity of the exposure conditions to obtain multiple sets of experimental data on the exposure intensity versus the pixel values of the test image. Based on the experimental data, the critical response value of the charge-coupled device (CCD) is determined; the critical response value is typically close to the full-well capacity threshold. The critical response value is divided into linear and nonlinear intervals, and then response models for the linear and nonlinear response intervals are fitted respectively. For example, the fitted photoelectric response model can be expressed as: In the above formula, S is the output signal of the charge-coupled device (i.e., the pixel value of the test image), I is the input signal (i.e., the exposure intensity), g is the slope obtained by linear fitting, and g is usually proportional to the system gain. S0 is the baseline signal in the absence of light, and I... c The critical value of the fitted response. For the sigmoid function, I M denoted as saturated light intensity, k is the transition slope obtained from the fitting, and k is used to reflect the steepness of the nonlinear interval curve. c is the offset position of the center point.
[0035] When the light intensity at a point is between 0 and the critical response value, an upward linear response interval function can be used. When the light intensity at a point is greater than or equal to the critical response value, a downward nonlinear response interval function can be used.
[0036] It's important to understand that, in addition to the sigmoid function, other curve functions, such as polynomial fitting, can be used for nonlinear response interval functions. The appropriate function can be selected based on the specific needs, and will not be elaborated upon here.
[0037] By fitting a photoelectric response model that includes both linear and nonlinear response interval functions, the measured spatial map can be accurately reconstructed across the entire light intensity domain, resulting in an actual light intensity image that reflects the true incident light intensity value.
[0038] In some embodiments of the grayscale quantization method of the present invention, as shown in FIG3, the simulated light intensity values at multiple preset positions in the simulated spatial image are obtained, and the incident light intensity values at the measured positions corresponding to each preset position in the actual light intensity image are obtained, including: S411, obtaining the central region of the high transmittance region of the target mask; S412, determining the highlight region corresponding to the central region of the high transmittance region in the simulated spatial image; S413, obtaining the simulated light intensity value of the highlight region, and obtaining the incident light intensity value at the measured position corresponding to the highlight region in the actual light intensity image; S414, obtaining the central region of the low transmittance region of the target mask; S415, determining the low light region corresponding to the central region of the low transmittance region in the simulated spatial image; S416, obtaining the simulated light intensity value of the low light region, and obtaining the incident light intensity value at the measured position corresponding to the low light region in the actual light intensity image.
[0039] In this embodiment, the high transmittance region can be the exposed region in the target mask, with a transmittance greater than 90%. The low transmittance region can be the unexposed region in the target mask, with a transmittance less than 50%.
[0040] In this embodiment, the minimum distance between the center region of the high transmittance region and the edge of the high transmittance region can be much greater than the wavelength of the light beam in the photolithography process. Similarly, the minimum distance between the center region of the low transmittance region and the edge of the low transmittance region is also much greater than the wavelength of the light beam in the photolithography process. This allows the diffraction effect to be ignored in the center regions of both the high transmittance and low transmittance regions, thereby eliminating the diffraction effect and improving the accuracy of the benchmark.
[0041] When calibrating the simulated light intensity values, the high-light region corresponding to the center of the flat high-transmittance region and the low-light region corresponding to the center of the flat low-transmittance region are used as sampling areas. The simulated light intensity value i of the high-light region... f1The simulated light intensity value i can be used as the maximum light intensity of the entire simulated space image, and the simulated light intensity value of the low-light region. f0 The minimum light intensity can be used as the simulated light intensity value i for the entire simulated space image. f1 and i f0 As a benchmark, it can reflect the situation of the simulated spatial image from the entire light intensity domain.
[0042] Next, the incident light intensity value I corresponding to the highlight region is obtained from the actual light intensity image. f1 And the incident light intensity value I corresponding to the low-light region. f0 Since the simulated light intensity values are relatively uniform and consistent in the central regions of both the flat high-transmittance region and the flat low-transmittance region, the corresponding incident light intensity values can be quickly and accurately matched from the actual light intensity images.
[0043] By obtaining the incident light intensity values of the two regions, using the incident light intensity values to calibrate the simulated light intensity values, and then extending this to the entire simulated spatial image, the relative mapping relationship between the simulated light intensity values and the corrected light intensity values of the entire simulated spatial image can be obtained, thereby further improving the accuracy of the calibration and enhancing the robustness of the relative mapping relationship.
[0044] In practical use, it is also possible to obtain the simulated light intensity values of multiple highlight areas and low light areas, as well as the corresponding multiple incident light intensity values, and calculate the average value of each as a benchmark to further improve the accuracy of the benchmark.
[0045] In some embodiments of the grayscale measurement method of the present invention, as shown in FIG4, the simulated light intensity values at various locations of the simulated spatial image are calibrated using simulated light intensity values at multiple preset locations and incident light intensity values at corresponding measured locations to obtain a corrected light intensity image. This includes: S511, using the incident light intensity value at the measured location corresponding to the highlight area as the corrected light intensity value of the simulated light intensity value of the highlight area, and using the incident light intensity value at the measured location corresponding to the low light area as the corrected light intensity value of the simulated light intensity value of the low light area, and constructing a linear regression model of the corrected light intensity value with respect to the simulated light intensity value; S513, using the linear regression model to calibrate the simulated light intensity values at various locations of the simulated spatial image to the corresponding corrected light intensity values to obtain a corrected light intensity image.
[0046] In this embodiment, the linear regression model is used to correct the relationship between the light intensity value and the simulated light intensity value, and is used to calibrate the simulated light intensity value based on preset optical parameters using the incident light intensity value obtained based on actual optical parameters.
[0047] For example, based on a linear regression model, the corrected light intensity values at various points in the corrected light intensity image can be expressed as: In the above formula, i s,r(x, y) represents the simulated light intensity value at any point (x, y) in the simulated spatial image, i f1 I represents the simulated light intensity value of the highlight region in the simulated spatial image. f1 i represents the incident light intensity value corresponding to the highlight region in the actual light intensity image. f0 I represents the simulated light intensity value of the low-light region in the simulated spatial image. f0 This represents the incident light intensity value corresponding to the low-light region in the actual light intensity image. The above formula describes the mapping relationship between the simulated light intensity value and the corrected light intensity value across the entire simulated spatial image.
[0048] In some embodiments of the grayscale measurement method of the present invention, as shown in FIG5, spatial registration of the corrected light intensity image and the actual light intensity image includes: S611, converting the corrected light intensity image into a pixelated image corresponding to the actual light intensity image; S613, performing sub-pixel level spatial registration of the pixelated image and the actual light intensity image.
[0049] The simulated spatial image is based on an optical model of a continuous physical field. The corrected light intensity image contains continuous corrected light intensity values (obtained from simulated light intensity values), and the density of the corrected light intensity value dataset is much higher than the resolution of the actual light intensity image. In order to compare the corrected light intensity image with the pixelated actual light intensity image, it is necessary to pixelate the corrected light intensity image and then spatially register the pixelated image with the actual light intensity image.
[0050] In the process of pixelating the corrected light intensity image into a pixelated image, the corrected light intensity image can be divided into multiple local regions (e.g., rectangular blocks). Then, the pixel value of the equivalent pixel in each local region can be determined based on the corrected light intensity value at each location within that local region. For example, the corrected light intensity value at the center of the local region can be used as the pixel value of the equivalent pixel in that local region, or the average corrected light intensity value of the local region can be used as the pixel value of the equivalent pixel in that local region. The appropriate calculation method can be selected as needed.
[0051] In the process of spatially registering a pixelated image with an actual light intensity image, subpixel resampling can be used to perform pixel-level translation on the pixelated image. For example, subpixel-level translations with effective accuracy can be iterated, and a full-pixel integral approximation formula can be applied to each translation to obtain multiple pixelated images. Then, the one with the highest matching degree to the actual light intensity image is determined as the final simulated spatial image. For example, interpolation algorithms with dynamic Lanczos kernel function cutoff range, Lanczos-4 interpolation algorithms with fixed Lanczos kernel function cutoff range, and bicubic interpolation algorithms can also be used to achieve subpixel-level affine transformation of the pixelated image. Then, among the multiple pixel-translated pixelated images, the one with the best registration to the actual light intensity image is determined, obtaining the target light intensity value of each equivalent pixel in the pixelated image.
[0052] In some embodiments of the grayscale quantization method of the present invention, as shown in FIG6, converting the corrected light intensity image into a pixelated image corresponding to the actual light intensity image includes: S621, dividing the corrected light intensity image into multiple local regions, wherein each local region corresponds one-to-one with the pixels of the actual light intensity image; S623, performing regional integration calculation on the corrected light intensity value for each local region, and using the calculation result as the equivalent pixel value of the local region to obtain a pixelated image, wherein the corrected light intensity value is the simulated light intensity value after calibration.
[0053] In this embodiment, local integral approximation is used to determine the pixel values of each equivalent pixel in the pixelated image. Specifically, the corrected light intensity image can first be pixelated to obtain multiple local regions corresponding to the pixels of the measured spatial image. For example, if the measured spatial image contains 1000×2000 pixels, the corrected light intensity image can be segmented into 1000×2000 local regions. Then, an integral approximation is performed on the corrected light intensity values in each local region to quickly obtain the pixel values of the equivalent pixels in that local region. For example, the following integral approximation formula can be used for calculation: In the above formula, I s,g (x, y) is the pixel value of the equivalent pixel centered at (x, y) in the pixelated image, corresponding to the incident light intensity value of a pixel in the measured spatial image. Ω is the local region centered at (x, y) in the corrected light intensity image, i s,g (u, v) is the corrected light intensity value of the point with coordinates (u, v) in Ω, and w is the preset weighting function.
[0054] The pixel values of each equivalent pixel obtained through integral approximation can accurately reflect the overall corrected light intensity value of each local area, thereby improving the fidelity of the obtained pixelated image.
[0055] In some embodiments of the gray quantization method of the present invention, the integral approximation operation is based on the local intensity integral operation of a weighted Gaussian kernel function. Specifically, the preset weight function can use the following Gaussian kernel: In the above formula, r is the distance between coordinates (u, v) and coordinates (x, y). By fusing the Gaussian distribution, the accuracy of the integral approximation calculation can be improved, making the pixel values of each equivalent pixel in the pixelated image more accurately reflect the light intensity of its local area as a whole.
[0056] In some embodiments of the grayscale measurement method of the present invention, as shown in FIG7, sub-pixel-level spatial registration of the pixelated image and the actual light intensity image includes: S631, performing sub-pixel-level displacement on the pixelated image according to multiple translation amounts to obtain a shifted image corresponding to each translation amount; wherein, for each equivalent pixel in each shifted image, the target light intensity value of each equivalent pixel is obtained by interpolating the corrected light intensity value of the neighborhood region corresponding to the pixel center of each equivalent pixel in the pixelated image; S633, based on the correlation analysis of mutual information, selecting the target light intensity value of each equivalent pixel in each shifted image that has the highest correlation with the incident light intensity value of each pixel in the actual light intensity image as the target registration image.
[0057] In a pixelated image, the centers of equivalent pixels may be offset from those of pixels in the actual intensity image. Directly comparing the pixelated image with the actual intensity image can lead to inaccurate results. Generally, a shift of ±0.5 pixels in the center coordinates of local regions relative to the center of pixels in the actual intensity image can cause changes in the integration region, resulting in discrepancies of up to 10% to 30%. Therefore, after obtaining the pixelated image, spatial registration between the pixelated image and the actual intensity image is necessary.
[0058] The scheme of determining the simulated spatial image with the highest matching degree to the actual light intensity image by traversing subpixel-level displacements with effective precision would generate a huge amount of computation, resulting in low spatial registration efficiency in practical applications. In this embodiment, an interpolation algorithm with a dynamic Lanczos kernel function truncation range is used to perform subpixel-level affine transformation on the pixelated image to further improve spatial registration efficiency. Specifically, during the affine transformation, the initial pixelated image can be subpixel resampled using the following formula: In the above formula, I' s,g (a, b) are I in the pixelated image s,g (x, y) represents the pixel value after displacement (a, b), where (a, b) is the displacement of the sub-pixel resampling center relative to the original pixel center in the pixelated image. (i, j) represents the integer pixel coordinates within the neighborhood of the sub-pixel resampling center, where... , , where N is the cutoff range of the Lanczos kernel function. q a and q b These are the Lanczos-N kernel weights in the x-axis and y-axis directions, respectively. , The Lanczos kernel L(x) consists of the sinc function and the Lanczos window: in, .
[0059] In this embodiment, different displacement amounts (a, b) can be set to obtain multiple deformed pixelated images, which are then compared with experimental optical images. Finally, based on the mutual information evaluation of pixel values, the one that is best aligned with the actual light intensity image among the deformed pixelated images is selected as the final simulated spatial image.
[0060] By performing subpixel-level affine transformation based on interpolation on the initial pixelated image, grayscale errors (such as edge blurring and peak shift) caused by pixel-level misalignment can be significantly eliminated, detail matching accuracy can be improved, and spatial registration efficiency can be greatly increased. A deformed pixelated image that is best aligned with the actual light intensity image can be obtained quickly.
[0061] In some embodiments of the gray quantization method of the present invention, the interpolation operation is based on the dynamic adaptive interpolation operation of the Lanczos-N kernel. The truncation parameter N of the Lanczos-N kernel is dynamically adjusted according to the local spectral characteristics of the image. The local spectral characteristics of the image are estimated by DCT (Discrete Cosine Transform) or FFT (Fast Fourier Transform) or gradient analysis to determine the adjustment value of the truncation parameter N.
[0062] The dynamic value of the Lanczos kernel function cutoff range N depends on the local frequency feature f. In this embodiment, the value of N can be dynamically determined based on the local frequency feature f by estimation through DCT, FFT, or gradient analysis, thereby improving the accuracy of the affine transformation.
[0063] In this embodiment, a smaller N can be used in the high-frequency region to maintain edge sharpness. A smaller N can be used in the low-frequency region to suppress ringing effects. For example, the value of N can be determined using the following formula: In the above formula, f M and f m These are the maximum and minimum local frequencies in the initial pixelated image, respectively.
[0064] 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.
[0065] 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.
[0066] This invention also provides a computer program product 10, a computer-readable storage medium 20, and a computer device 30. FIG8 is a schematic diagram of a computer program product 10 according to an embodiment of the present invention, FIG9 is a schematic diagram of a computer-readable storage medium 20 according to an embodiment of the present invention, and FIG10 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 a processor 32, implements the steps of the grayscale measurement method for a masked image described above. The computer-readable storage medium 20 stores the computer program 11 thereon, which, when executed by the processor 32, implements the steps of the grayscale measurement method for a masked image described above. The computer device 30 may include a memory 31, a processor 32, and a computer program 11 stored in the memory 31 and running on the processor 32.
[0067] 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.
[0068] For the purposes of this embodiment, computer program product 10 is a related product that includes computer program 11.
[0069] 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.
[0070] 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.
[0071] 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.
[0072] 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.
[0073] 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 grayscale measurement method for a masked realistic image, characterized in that, include: Simulate the target mask to obtain a simulated spatial image; acquire the measured spatial image of the target mask acquired by the image acquisition device; Using a photoelectric response model, the measured pixel values of the measured spatial image are restored to the incident light intensity values to obtain an actual light intensity image. The photoelectric response model is configured to characterize the photoelectric response characteristics of the image acquisition device. The simulated light intensity values at multiple preset positions in the simulated space image are obtained, and the incident light intensity values at the measured positions corresponding to each preset position in the actual light intensity image are obtained. Using the simulated light intensity values at multiple preset locations and the incident light intensity values at corresponding measured locations, the simulated light intensity values at various points in the simulated spatial image are calibrated to obtain a corrected light intensity image; the corrected light intensity image and the actual light intensity image are then spatially registered.
2. The gray measurement method according to claim 1, characterized in that, The steps for constructing the photoelectric response model include: acquiring multiple test images obtained by the image acquisition device under various exposure conditions, and obtaining multiple sets of experimental data on exposure intensity and corresponding pixel values of the test images; determining the response critical value of the image acquisition device by analyzing the multiple sets of experimental data; wherein, when the exposure intensity is less than the response critical value, the exposure intensity and the corresponding pixel value of the test image have a linear relationship; and when the exposure intensity is greater than or equal to the response critical value, the exposure intensity and the corresponding pixel value of the test image have a non-linear relationship; using the multiple sets of experimental data, with the response critical value as the interval boundary point, fitting a linear relationship interval function and a non-linear relationship interval function regarding the exposure intensity and the corresponding pixel value of the test image; and constructing the photoelectric response model based on the linear relationship interval function and the non-linear relationship interval function.
3. The gray measurement method according to claim 1, characterized in that, The step of obtaining simulated light intensity values at multiple preset positions in the simulated spatial image and obtaining the incident light intensity values at measured positions corresponding to each preset position in the actual light intensity image includes: obtaining the central region of the high transmittance region of the target mask; determining a highlight region in the simulated spatial image corresponding to the central region of the high transmittance region; obtaining the simulated light intensity value of the highlight region and obtaining the incident light intensity value at the measured position corresponding to the highlight region in the actual light intensity image; obtaining the central region of the low transmittance region of the target mask; determining a low-light region in the simulated spatial image corresponding to the central region of the low transmittance region; obtaining the simulated light intensity value of the low-light region and obtaining the incident light intensity value at the measured position corresponding to the low-light region in the actual light intensity image.
4. The gray measurement method according to claim 3, characterized in that, The step of calibrating the simulated light intensity values at various locations in the simulated spatial image using the simulated light intensity values at multiple preset locations and the incident light intensity values at corresponding measured locations to obtain a corrected light intensity image includes: using the incident light intensity value at the measured location corresponding to the highlight region as the corrected light intensity value of the simulated light intensity value of the highlight region, using the incident light intensity value at the measured location corresponding to the low-light region as the corrected light intensity value of the simulated light intensity value of the low-light region, constructing a linear regression model of the corrected light intensity value with respect to the simulated light intensity value; and using the linear regression model to calibrate the simulated light intensity values at various locations in the simulated spatial image to correspond to the corrected light intensity values to obtain the corrected light intensity image.
5. The gray measurement method according to claim 1, characterized in that, The step of spatially registering the corrected light intensity image and the actual light intensity image includes: converting the corrected light intensity image into a pixelated image corresponding to the actual light intensity image; and performing sub-pixel-level spatial registration between the pixelated image and the actual light intensity image.
6. The gray measurement method according to claim 5, characterized in that, The step of converting the corrected light intensity image into a pixelated image corresponding to the actual light intensity image includes: dividing the corrected light intensity image into multiple local regions, each local region corresponding one-to-one with the pixels of the actual light intensity image; performing regional integration calculations on the corrected light intensity value for each local region, and using the calculation results as the equivalent pixel value of the local region to obtain the pixelated image, wherein the corrected light intensity value is the simulated light intensity value after calibration.
7. The gray measurement method according to claim 6, characterized in that, The subpixel-level spatial registration of the pixelated image and the actual light intensity image includes: performing subpixel-level displacement on the pixelated image according to multiple translation amounts to obtain a shifted image corresponding to each translation amount; wherein, for each equivalent pixel in each shifted image, the target light intensity value of each equivalent pixel is obtained by interpolating the corrected light intensity value of the neighborhood region corresponding to the pixel center of each equivalent pixel in the pixelated image; based on the correlation analysis of mutual information, the target light intensity value of each equivalent pixel in each shifted image that has the highest correlation with the incident light intensity value of each pixel in the actual light intensity image is selected as the target registration image.
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 grayscale measurement method for a masked real image as described in any one of claims 1 to 7.
9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the grayscale measurement method for the masked real image 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 grayscale measurement method for the masked simulating image according to any one of claims 1 to 7.