Pattern wafer surface three-dimensional shape reconstruction method

Abstract

The invention discloses a pattern wafer surface three-dimensional morphology reconstruction method, which comprises the following steps of: 1, constructing a small depth-of-field microscopic imaging system, and scanning the surface of a wafer along an optical axis direction to obtain a zoom image sequence; 2, establishing an image sequence space and a color vector space, calculating a pixel color gradient and a dispersion characteristic as a focusing evaluation basis, fitting a Gaussian distribution curve, obtaining a focusing evaluation curve with high steepness and an accurate center position according to a Bayesian prior distribution and posterior distribution principle, and performing focusing evaluation on the focusing evaluation curve; combining the curve peak value with the scanning interval to obtain a wafer surface initial depth map; and 3, performing filtering optimization on the initial depth map, and outputting a final three-dimensional reconstruction result. The method can effectively solve the problem that low-texture and low-contrast area definition evaluation is not accurate in a traditional focusing evaluation method, can improve the focusing peak searching speed and precision, and is suitable for high-precision microscopic three-dimensional reconstruction work of the wafer surface.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and particularly to a three-dimensional reconstruction method for the absolute depth of the surface of a wafer to be measured under a microscope with high precision. Background Art

[0002] With the continuous refinement of semiconductor manufacturing processes, the high-precision detection of the three-dimensional topography of the wafer surface has become a key link to ensure the yield and performance of chips. Traditional three-dimensional reconstruction methods are mostly based on the focus stacking technology. By collecting a sequence of zoom images along the optical axis direction through a microscopic system and using a focus evaluation function to determine the best focus position of each pixel, the surface height information is reconstructed. However, the existing technologies have significant deficiencies in the following aspects: Traditional methods usually convert color images into grayscale images for focus evaluation, resulting in the loss of differential information between color channels, thereby reducing the detection sensitivity and even causing misjudgment of key features; In addition, existing focus evaluation functions (such as gradient magnitude, Laplacian energy, etc.) are mostly based on single-channel grayscale information, are susceptible to noise interference, and the evaluation curves often show multi-peak or wide-peak phenomena, resulting in blurred peak positioning; Limitations in depth map optimization, the initial depth map often introduces errors due to noise or outliers, and traditional filtering methods (such as mean filtering) are prone to blur edge details while smoothing noise, and it is difficult to balance the smoothness of the surface flat area and the fidelity of micro-nano structures. For example, the step height or microscopic scratches of metal wires on the wafer surface may be distorted due to excessive filtering.

[0003] In recent years, although there have been research attempts to combine multi-color channel information or introduce statistical models to optimize focus evaluation, their methods are mostly limited to simple weighted fusion or linear assumptions and have not effectively solved the non-linear correlation problem of color feature information. In addition, for the characteristics of high reflectivity and low contrast of the wafer surface, the existing technologies still lack robust anti-interference capabilities and adaptive optimization mechanisms. Summary of the Invention

[0004] The present invention is to solve the above-mentioned deficiencies of the existing technologies, and proposes a method for reconstructing the three-dimensional topography of the surface of a patterned wafer, in order to be able to make full use of color information, improve the accuracy and speed of focus evaluation, avoid information loss caused by traditional grayscale conversion, improve the peak positioning speed and accuracy, thereby realizing high-fidelity optimization of the depth map for three-dimensional reconstruction, and effectively retaining the micro-structure details of the wafer surface to meet the stringent detection requirements for the high-reflection and low-contrast surface of the wafer surface in semiconductor manufacturing.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] The characteristics of a method for reconstructing the three-dimensional topography of the surface of a patterned wafer according to the present invention are that it is applied to a scenario composed of a wafer to be measured, a stage, a stepping motor, a CCD camera, a microscope, and a coaxial light source, and includes the following steps:

[0007] Step 1: Collect a set of zoom image sequences on the surface of the wafer to be measured , ,…, ,…, , where represents the th surface image of the wafer to be measured, represents the total number of images, represents the width of a single image, is the height of a single image;

[0008] Step 2: Arrange , ,…, ,…, in the order of acquisition along the optical axis direction; Take the first pixel point in the upper left corner of the first image as the origin of the image sequence space coordinate system, and take the two pixel coordinate axes of and the vertical axis parallel to the optical axis and passing through the origin as the three coordinate axes of the image sequence space coordinate system, thereby establishing the image sequence space; Denote any pixel point in the image sequence space as , where , , ;

[0009] Step 3: Based on the image sequence space and the color vector space, calculate the color gradient evaluation value and the dispersion evaluation value of all pixels, thereby obtaining the image gradient matrix and the image dispersion matrix , where is the color gradient evaluation value of the pixel point in the image sequence space, is the dispersion evaluation value of the pixel point in the color vector space;

[0010] Step 4: According to the image gradient matrix and the image dispersion matrix, obtain the normalized gradient set and the normalized dispersion set of each pixel point, and calculate the Gaussian posterior distribution function of each pixel point, and extract the function center position as the best focus position of each pixel point, thereby obtaining the initial depth map of the surface of the wafer to be measured , where is the depth value of the pixel point ;

[0011] Step 5: Perform median filtering and guided filtering on the initial depth map in turn to obtain the optimized depth map , which is used to realize the high-precision three-dimensional reconstruction of the surface of the wafer to be measured, where is a pixel point The optimized depth value

[0012] Another feature of the 3D topography reconstruction method for the surface of a patterned wafer according to the present invention is that step 3 includes:

[0013] Step 3.1: Use the black pixel point as the origin of the color vector space coordinate system, and use the red channel, green channel, and blue channel as the three base coordinate axes of the color vector space coordinate system, thereby constructing the color vector space;

[0014] Step 3.2: Use Equation (1) to obtain the mapping relationship from the image sequence space to the color vector space:

[0015] (1)

[0016] In Equation (1), represents the mapping, represents the vector point corresponding to the pixel point in the color vector space, , and respectively represent the pixel points on the components on the red channel, green channel, and blue channel;

[0017] Step 3.3: Use Equation (2) to obtain the color vector of the pixel point :

[0018] (2)

[0019] In Equation (2), , , respectively represent the unit vectors of the three base coordinate axes;

[0020] Step 3.4: In the color vector space, use Equation (3) and Equation (4) to calculate the horizontal direction gradient and the vertical direction gradient of the pixel point on , thereby obtaining the color partial derivative matrix of the pixel point :

[0021] (3)

[0022] (4)

[0023] Use Equation (5) to obtain the pixel point Color vector space correlation matrix :

[0024] (5)

[0025] Step 3.5: Calculate the color vector space correlation matrix The maximum eigenvalue and use it as the color gradient evaluation value of the pixel point in the image sequence space , so as to traverse all pixel points in the image sequence space and obtain the image gradient matrix ;

[0026] Step 3.6: With the pixel point as the center, draw a local window with a size of on . The th neighborhood pixel of the pixel point in the local window is denoted as the pixel point , and the color vector of the pixel point is denoted as ; where , is the length of the local window;

[0027] Step 3.7: Use Equation (6) to calculate the dispersion value of the pixel point and the pixel point , so as to obtain the dispersion values of the pixel point and all neighborhood pixels in the local window, and form the color difference set of the pixel point ;

[0028] (6)

[0029] In Equation (6), is the Euclidean distance between the color vector and the color vector ; is the color fluctuation amount of the pixel point , and there is:

[0030] (7)

[0031] In Equation (7), , , respectively represent the components of the surrounding pixel points of the pixel point on the red channel, green channel, and blue channel;

[0032] Step 3.8: Obtain the pixel dispersion evaluation value in the color vector space , thereby traversing all pixel points in the image sequence space and obtaining the image dispersion matrix ;

[0033] (8)

[0034] In formula (8), represents variance.

[0035] Furthermore, the said step 4 includes:

[0036] Step 4.1: Take the data points of the pixel on the third dimension of the image gradient matrix to form the gradient set of the pixel , and normalize the data points in to obtain the normalized gradient set of the pixel , thereby fitting to the gradient Gaussian function of the pixel , where represents the normalized color gradient evaluation value of the pixel ;

[0037] (9)

[0038] In formula (9), represents the mean value of the gradient Gaussian function , that is, the center position of the gradient Gaussian curve, represents the standard deviation of the gradient Gaussian function ;

[0039] Step 4.2: Take the data points of the pixel on the third dimension of the image dispersion matrix to form the dispersion set of the pixel , and normalize the data points in to obtain the normalized dispersion set of the pixel , thereby fitting to the dispersion Gaussian function of the pixel , where represents the normalized dispersion evaluation value of the pixel ;

[0040] (10)

[0041] In formula (10), represents the mean value of the dispersion Gaussian function , represents the dispersion Gaussian function 's standard deviation;

[0042] Step 4.3: Take as the prior function, take as the likelihood function, and according to the Bayesian statistical principle, use formula (11) to calculate the Gaussian posterior distribution function of the pixel point : :

[0043] (11)

[0044] In formula (11), represents the marginal likelihood function of the pixel point ;

[0045] Step 4.4: Use formulas (12) and (13) to calculate the mean value and variance of the Gaussian posterior distribution function respectively, and take as the best focus position of the pixel point :

[0046] (12)

[0047] (13)

[0048] Step 4.5: Use formula (14) to calculate the depth value of the pixel point :

[0049] (14)

[0050] In formula (21), represents 's scanning lower limit, represents the scanning step size;

[0051] Step 4.6: Traverse all pixels in the image according to the process of Step 4.1 to Step 4.5, obtain the depth value of each pixel point, and thus obtain the initial depth map of the surface of the wafer to be measured .

[0052] An electronic device according to the present invention includes a memory and a processor, characterized in that the memory is used to store a program for supporting the processor to execute the three-dimensional topography reconstruction method of the pattern wafer surface according to any one of claims 1-3, and the processor is configured to execute the program stored in the memory.

[0053] A computer-readable storage medium according to the present invention, characterized in that when the computer program stored on the computer-readable storage medium is run by a processor, it executes the steps of the three-dimensional topography reconstruction method of the pattern wafer surface according to any one of claims 1-3.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0055] 1. The present invention establishes an image sequence space and a color vector space, calculates the pixel color gradient and dispersion characteristics as the basis for sharpness evaluation, makes full use of the color channel correlation, improves the accuracy and stability of the aggregation evaluation link, and effectively solves the problem that the traditional focus evaluation algorithm has inaccurate evaluation of low-texture and low-contrast regions, thus providing guarantee for the accuracy of stacked focus three-dimensional reconstruction;

[0056] 2. The present invention proposes to use the Gaussian curve of the pixel color gradient as the prior distribution function and the Gaussian curve of the dispersion characteristic as the likelihood function, and obtains a focus evaluation curve with more accurate center position and higher steepness through the Bayesian statistical principle, avoiding the inherent error caused by interpolation fitting in the traditional peak search link, improving the accuracy of depth estimation, and thus realizing high-precision three-dimensional reconstruction of the wafer surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is a schematic diagram of obtaining the depth of the surface of the wafer to be measured according to the present invention;

[0058] Figure 2 is a flowchart of collecting a color image sequence according to the present invention;

[0059] Figure 3 is the zoomed image sequence processed according to the present invention;

[0060] Figure 4 is a focus evaluation curve graph of the image processed according to the present invention;

[0061] Figure 5 is the surface of the wafer reconstructed according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] In this embodiment, a method for three-dimensional topography reconstruction of a pattern wafer surface has a principle flow as Figure 1 shown, and is applied to a scenario composed of a wafer to be measured, a stage, a stepping motor, a CCD camera, a microscope, and a coaxial light source, and includes the following steps:

[0063] Step 1: Collect a set of zoom image sequences on the surface of the wafer to be measured , ,…, ,…, , where represents the th surface image of the wafer to be measured, represents the total number of images, represents the width of a single image, is the height of a single image;

[0064] Step 1.1: The process of collecting the wafer image sequence is as Figure 2 shown. The CCD camera and the high-precision stepping motor work together. Through the coaxial light source, a uniform illumination beam is generated. The beam forms a parallel optical path through the collimating mirror and is projected onto the surface of the object to be measured by the microscopic objective lens after reflection at the beam splitter;

[0065] Step 1.2: Control the stepping motor to continuously move the microscopic objective lens closer to the object to be measured, and monitor the global average gradient value of the CCD image in real time. When the continuous decrease amplitude of the gradient value for 5 consecutive frames ≤ 0.1%, it is determined that all pixels enter the defocus blur state, and record the Z-axis coordinate at this time as . Then, reverse-control the stepping motor to gradually move the microscopic objective lens away from the object to be measured. When the gradient evaluation value of more than 95% of the pixels in the image drops to 10% of the peak value, record the Z-axis coordinate at this time as .

[0066] Step 1.3: According to the total scanning distance , calculate the displacement step size according to the formula , and set the constraint conditions; Multi-scale scanning control: Starting from , the stepping motor moves along the Z-axis by step size, and triggers the CCD camera to take pictures every time a new position is reached, generating an image sequence , ,…, ,…, , as Figure 3 shown.

[0067] Step 2: Establish the image sequence space and the color vector space, and calculate the color gradient evaluation value and the dispersion evaluation value of all pixels, so as to obtain the image gradient matrix and the image dispersion matrix:

[0068] Step 2.1: Arrange , ,…, ,…, in the order of acquisition along the optical axis direction. Each image is parallel to its two adjacent images and has the same interval; The first image The first pixel at the upper left corner is used as the origin of the spatial coordinate system of the image sequence, and the two pixel coordinate axes of and the vertical axis parallel to the optical axis and passing through the origin are used as the three coordinate axes of the spatial coordinate system of the image sequence, thereby establishing the spatial coordinate system of the image sequence. Let any pixel point in the spatial coordinate system of the image sequence be denoted as where, .

[0069] Step 2.2: Using the black pixel as the origin of the color vector space coordinate system, and using the red channel, green channel, and blue channel as the three base coordinate axes of the color vector space coordinate system, thereby constructing the color vector space;

[0070] The mapping relationship from the spatial coordinate system of the image sequence to the color vector space is obtained by using Equation (15):

[0071] (15)

[0072] In Equation (15), represents the mapping, represents the vector point corresponding to the pixel point in the color vector space, , and respectively represent the components of the pixel point on the red channel, green channel, and blue channel.

[0073] The color vector of the pixel point is obtained by using Equation (16):

[0074] (16)

[0075] In Equation (16), , , respectively represent the unit vectors of the three base coordinate axes.

[0076] Step 2.3: In the color vector space, calculate the horizontal gradient and the vertical gradient of the pixel point on using Equations (17) and (18), thereby obtaining the color partial derivative matrix of the pixel point :

[0077] (17)

[0078] ​ (18)

[0079] Obtain the pixel point by using Equation (19) of the color vector space correlation matrix :

[0080] (19)

[0081] The spatial correlation matrix characterizes the correlation of the three channels in the RGB space. The eigenvectors of the matrix reflect the changes of the image function in the RGB space. The amplitude of the eigenvector represents the magnitude of the change rate, and the direction of the eigenvector represents the change direction. The direction and amplitude of the maximum change of the color image function are represented by the eigenvector corresponding to the largest eigenvalue of the matrix; calculate the color vector space correlation matrix of the largest eigenvalue , and use it as the color gradient evaluation value of the pixel point , so as to traverse all pixel points in the image sequence space and obtain the image gradient matrix . .

[0082] Step 2.4: Centered on the pixel point , draw a local window with a size of on . Denote the th neighborhood pixel of the pixel point in the local window as the pixel point , and denote the color vector of the pixel point as ; where , is the length of the local window;

[0083] Calculate the dispersion value of the pixel point and the pixel point by using Equation (20), so as to obtain the dispersion values of the pixel point and all neighborhood pixels in the local window, and form the color difference set of the pixel point ;

[0084] (20)

[0085] In Equation (20), is the Euclidean distance between the color vector and the color vector . The Euclidean distance can intuitively represent the similarity between two vectors. The larger its value, the greater the difference between the two vectors, indicating that the pixel point and the pixel point The more obvious the color difference is; is the pixel point of the color fluctuation amount, and there is:

[0086] (21)

[0087] In formula (21), is the pixel point of the three-dimensional coordinates of the surrounding pixels.

[0088] Using formula (22) to obtain the dispersion evaluation value of the pixel point in the color vector space, thereby traversing all pixel points in the image sequence space and obtaining the image dispersion matrix ;

[0089] (22)

[0090] In formula (22), represents variance; calculate the variance and sum of the color difference set and multiply the two as the dispersion evaluation value of the pixel point in the color vector space, which can accurately evaluate the focusing degree of the pixel.

[0091] Step 3: According to the image gradient matrix and the image dispersion matrix, obtain the normalized gradient set and the normalized dispersion set of each pixel point, calculate the Gaussian posterior distribution function of each pixel point, and extract the center position of the function as the best focusing position of each pixel point, thereby obtaining the initial depth map of the surface of the wafer to be measured:

[0092] Step 3.1: Take the pixel point in the image gradient matrix to form a gradient set from the data points in the third dimension, and normalize the data points in to obtain the normalized gradient set , thereby using formula (23) to fit into a gradient Gaussian function , where represents the normalized color gradient evaluation value of the pixel point ;

[0093] (23)

[0094] In formula (23), represents the mean value of the gradient Gaussian function, that is, the center position of the gradient Gaussian curve, represents the standard deviation of the gradient Gaussian function, reflecting the width of the Gaussian curve.

[0095] Step 3.2: Select pixel points On the third dimension of the image dispersion matrix The data points form a dispersion set , and Normalize the data points in to obtain a normalized dispersion set , and then fit to the dispersion Gaussian function using Equation (24) , where represents the normalized dispersion evaluation value of the pixel point ; Normalized dispersion evaluation value;

[0096] (24)

[0097] In Equation (24), represents the mean of the dispersion Gaussian function, represents the standard deviation of the dispersion Gaussian function.

[0098] Step 3.3: Use as the prior function, and as the likelihood function. According to the Bayesian statistical principle, calculate the Gaussian posterior distribution function using Equation (11) , as shown in Figure 4 ; Both the prior and likelihood functions are Gaussian distributions, and the posterior distribution obtained according to the Bayesian formula is also a Gaussian distribution, and the posterior distribution curve will be steeper, improving the response speed and resolution of the focus evaluation;

[0099] (25)

[0100] In Equation (25), represents the marginal likelihood function.

[0101] Calculate the mean and variance of the Gaussian posterior distribution function using Equations (12) and (13) respectively, and use as the best focus position of the pixel point :

[0102] (28)

[0103] (29)

[0104] Step 3.4: Combine the best focus position of the pixel point with the scanning interval , and calculate the depth value of the pixel point using Equation (28) :

[0105] (28)

[0106] In formula (28), represents the lower limit of the scan, and represents the step size of the scan.

[0107] Step 3.5: Traverse all pixels in the image according to the process of Step 3.1 to Step 3.4, obtain the depth value of each pixel point, and thus obtain the depth map of the surface of the wafer to be measured. ;

[0108] Step 4: Perform median filtering and guided filtering on the initial depth map in sequence to obtain an optimized depth map , which is used to achieve high-precision 3D reconstruction of the surface of the wafer to be measured:

[0109] Step 4.1: To suppress noise interference, perform median filtering on the initial depth map , set the sliding window size to 3×3, remove outliers and smooth local fluctuations, and obtain the depth map :

[0110] Step 4.2: Perform guided filtering on to obtain the finally optimized depth map ;

[0111] Step 4.3: As shown in Figure 5 , convert the optimized depth map into 3D point cloud data, combine the calibration parameters of the microscopic system, map it to the physical coordinate system, and generate a 3D topography model of the surface of the object to be measured through a surface fitting algorithm, and output the final 3D reconstruction result.

[0112] In this embodiment, an electronic device includes a memory and a processor. The memory is used to store a program that supports the processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0113] In this embodiment, a computer-readable storage medium stores a computer program on the computer-readable storage medium. When the computer program is run by a processor, it executes the steps of the above method.

Claims

1. A method for reconstructing the three-dimensional topography of a patterned wafer surface, characterized in that, It is applied to a scenario composed of a wafer to be measured, a stage, a stepping motor, a CCD camera, a microscope, and a coaxial light source, and includes the following steps: Step 1: Collect a set of zoomed image sequences on the surface of the wafer to be measured , ,…, ,…, , where represents the th image on the surface of the wafer to be measured, represents the total number of images, represents the width of a single image, is the height of a single image; Step 2: Arrange , , …, , …, in the order of acquisition along the optical axis direction; take the first pixel point in the upper left corner of the first image as the origin of the image sequence space coordinate system, and take the two pixel coordinate axes of and the vertical axis parallel to the optical axis and passing through the origin as the three coordinate axes of the image sequence space coordinate system, so as to establish the image sequence space; denote any pixel point in the image sequence space as , where , , ;​​​​​​​​​​​​​​​​​​​​ Step 3: Based on the image sequence space and the color vector space, calculate the color gradient evaluation value and the dispersion evaluation value of all pixels, so as to obtain the image gradient matrix and the image dispersion matrix , where is the color gradient evaluation value of the pixel point in the image sequence space, and is the dispersion evaluation value of the pixel point in the color vector space; Step 4: According to the image gradient matrix and the image dispersion matrix, obtain the normalized gradient set and the normalized dispersion set of each pixel, and calculate the Gaussian posterior distribution function of each pixel, so as to extract the function center position as the best focus position of each pixel, thereby obtaining the initial depth map of the surface of the wafer to be measured , where is the depth value of pixel ; Step 5: For the initial depth map perform median filtering and guided filtering in sequence to obtain an optimized depth map , which is used to achieve high-precision three-dimensional reconstruction of the surface of the wafer to be measured. Among them, is the pixel point and is the optimized depth value.

2. The three-dimensional topography reconstruction method for the surface of a patterned wafer according to claim 1, wherein Step 3 includes: Step 3.1: Taking the black pixel points as the origin of the color vector space coordinate system, and using the red channel, the green channel, and the blue channel as the three base coordinate axes of the color vector space coordinate system, thereby constructing the color vector space; Step 3.2: Using Equation (1) to obtain the mapping relationship from the image sequence space to the color vector space: (1) In formula (1), represents a mapping, represents the vector point corresponding to the pixel point in the color vector space, , and respectively represent the pixel points on the components on the red channel, green channel, and blue channel; Step 3.3: Obtain the color vector of the pixel point using Equation (2) : (2) In formula (2), , , respectively represent the unit vectors of the three base coordinate axes; Step 3.4: In the color vector space, calculate the pixel points on horizontal gradient of and vertical gradient of , so as to obtain the color partial derivative matrix of the pixel point : (3) (4) Obtain the pixel point using Equation (5) correlation matrix of the color vector space : (5) Step 3.5: Calculate the correlation matrix of the color vector space of the maximum eigenvalue , and used as the pixel in the image sequence space of the color gradient evaluation value , so as to traverse all pixels in the image sequence space, to obtain the image gradient matrix ; Step 3.6: With the pixel point as the center, draw a local window with a size of on . The th neighborhood pixel of the pixel points within the local window is denoted as pixel point , and the color vector of pixel point is denoted as ; where , and is the length of the local window. Step 3.7: Calculate the pixel point and the pixel point chromatic dispersion value , so as to obtain the chromatic dispersion values of the pixel point and all neighboring pixels within the local window, and form the color difference set of the pixel point ;​ (6) In formula (6), is the Euclidean distance between the color vectors and the color vector ; is the color fluctuation amount of the pixel , and there is: (7) In formula (7), , , respectively represent the components of the surrounding pixels of the pixel point on the red channel, green channel, and blue channel; Step 3.8: Obtain the pixel dispersion evaluation value in the color vector space , thereby traversing all pixel points in the image sequence space and obtaining the image dispersion matrix ; (8) In formula (8), represents the variance.

3. A method for reconstructing the three-dimensional topography of a patterned wafer surface according to claim 2, wherein, Step 4 includes: Step 4.1: Obtain pixel points The data points on the third dimension of the image gradient matrix constitute the gradient set of the pixel points , and normalize the data points in , to obtain the normalized gradient set of the pixel points . Then, fit the data in to the gradient Gaussian function of the pixel points using Equation (9), where represents the normalized color gradient evaluation value of the pixel points . Here, represents the normalized color gradient evaluation value of the pixel points (9) In formula (9), represents the mean value of the gradient Gaussian function , that is, the central position of the gradient Gaussian curve, represents the standard deviation of the gradient Gaussian function ; Step 4.2: Obtain pixel points The data points on the third dimension of the image dispersion matrix constitute the dispersion set of pixel points , and normalize the data points in to obtain the normalized dispersion set of pixel points , thereby fitting the data in into the dispersion Gaussian function of pixel points using Equation (10), where represents the normalized dispersion evaluation value of pixel points ; Among them, represents the pixel point ; (10) In formula (10), represents the mean value of the dispersion Gaussian function , and represents the standard deviation of the dispersion Gaussian function . Step 4.3: Take as the prior function, and take as the likelihood function. According to the Bayesian statistical principle, use Equation (11) to calculate the Gaussian posterior distribution function of the pixel point : : (11) In Equation (11), represents the marginal likelihood function of the pixel point ; Step 4.4: Calculate the mean value and variance of the Gaussian posterior distribution function using equations (12) and (13) respectively, and take as the best focusing position of the pixel point : ​ (12) (13) Step 4.5: Calculate the depth value of the pixel point using Equation (14) : (14) In formula (21), represents the lower limit of the scan, represents the step size of the scan; Step 4.6: Traverse all the pixels in the image according to the process from Step 4.1 to Step 4.5, obtain the depth value of each pixel point, and thus obtain the initial depth map of the surface of the wafer to be measured .

4. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor to execute the method for reconstructing the three-dimensional topography of the surface of the patterned wafer according to any one of claims 1-3, and the processor is configured to execute the program stored in the memory.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it executes the steps of the method for reconstructing the three-dimensional topography of the surface of the patterned wafer according to any one of claims 1-3.

Images