Aperture-related confocal microscopic measurement system and its three-dimensional reconstruction method

Abstract

The aperture-related confocal microscopic measurement system and its three-dimensional reconstruction method of the present invention include: a light source, a first collimating lens, a first polarizer, a polarization beam splitter, a turntable, a tube lens, a quarter-wave plate, an objective lens, a second polarizer, an imaging lens, and a detector. The present invention proposes a three-dimensional reconstruction algorithm for an aperture-related confocal microscope. Based on Gaussian fitting of the weighted least squares method and its iterative processing method, it can quickly and accurately fit the height position of the surface points of the sample. While retaining the fast scanning characteristics of the turntable, by subtracting the defocus interference from the composite image, the influence of crosstalk between apertures is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the field of confocal microscopy measurement, and particularly to an aperture-related confocal microscopy measurement system and a three-dimensional reconstruction method thereof. Background Art

[0002] A confocal microscopy system restricts the illumination range through a pinhole and blocks out-of-focus light simultaneously, thereby achieving high-contrast imaging of a single point. By scanning point by point and line by line, a high-definition full-frame image can be obtained. This technology is not easily interfered by out-of-focus light, enabling it to have a powerful optical sectioning ability and allowing for three-dimensional imaging of thick objects. Therefore, it is often used in three-dimensional detection applications and can precisely measure the surface topography of samples.

[0003] Based on the above principle, a spinning disk confocal microscope sets a plurality of pinholes, slits or various mask patterns on a disk to achieve a similar confocal effect, and then rotates the disk to perform multi-point or line scanning of the sample surface, thereby achieving fast surface imaging. It is a confocal type specialized in speed. However, out-of-focus light can pass through adjacent apertures, causing the problem of crosstalk between apertures.

[0004] While the aperture-related confocal microscopy system retains the fast scanning characteristics of the spinning disk, the influence of crosstalk between apertures is greatly reduced by subtracting the out-of-focus interference (a wide-field image of a certain intensity) from the composite image. The axial response curve of any point on the sample to the confocal microscopy measurement system is approximately Gaussian distributed. The core of the three-dimensional reconstruction algorithm lies in the fitting of the axial response curve, and the peak of the curve is the height position of that point.

[0005] Existing Gaussian fitting algorithms include:

[0006] (1) The least squares fitting method transformed into a second-order polynomial. This method is fast, but has poor accuracy and is easily interfered by noise, especially noise at low-value points.

[0007] (2) The non-linear Gaussian fitting method, generally the LM algorithm, has high accuracy and strong anti-interference ability, but is slow. Summary of the Invention

[0008] In order to overcome the deficiencies of the prior art, the present invention provides an aperture-related confocal microscopy measurement system and a three-dimensional reconstruction method thereof. Through a scientific system architecture design, it can achieve a similar measurement speed, perform data fitting processing on the images obtained by the aperture-related confocal microscope, and adopt a three-dimensional reconstruction algorithm. Based on the Gaussian fitting of the weighted least squares method and its iterative processing method, it can quickly and accurately fit the height position of the points on the sample surface. While retaining the fast scanning characteristics of the spinning disk, the influence of crosstalk between apertures is greatly reduced by subtracting the out-of-focus interference (i.e., a wide-field image of a certain intensity) from the composite image.

[0009] To solve the above technical problems, the present invention provides an aperture-related confocal microscopic measurement system and a three-dimensional reconstruction method thereof, wherein:

[0010] The aperture-related confocal microscopic measurement system includes:

[0011] A light source, a first collimating lens, a first polarizer, a polarization beam splitter, a turntable, a tube lens, a quarter-wave plate, an objective lens, a second polarizer, an imaging lens, and a detector;

[0012] The first collimating lens is disposed below the light source, the first polarizer is disposed below the first collimating lens, the polarization beam splitter is disposed below the first polarizer, the turntable is disposed below the polarization beam splitter, the tube lens is disposed below the turntable, the quarter-wave plate is disposed below the tube lens, the objective lens is disposed below the quarter-wave plate, and a sample is placed below the objective lens;

[0013] As an example, the sample is placed on a displacement stage.

[0014] The second polarizer is disposed on the left side of the polarization beam splitter, the imaging lens is disposed on the left side of the second polarizer, and the detector is disposed on the left side of the imaging lens; and it is ensured that the optical axis of the second polarizer is orthogonal to the optical axis of the first polarizer;

[0015] The light source is used to emit illumination light. After the illumination light is collimated by the collimating lens, it passes through the first polarizer to generate incident linearly polarized light; the incident linearly polarized light is transmitted at the polarization beam splitter, and the transmitted light is modulated by the turntable. The modulated light is focused on the rear focal plane of the objective lens by the tube lens again. During this period, the illumination light passes through the quarter-wave plate and becomes circularly polarized light; the circularly polarized light is projected onto the sample by the objective lens;

[0016] The reflected light generated on the sample returns along the original path. At the quarter-wave plate, it becomes reflected linearly polarized light with a vibration direction orthogonal to the incident light, and then is focused on the turntable by the tube lens. The turntable demodulates the reflected linearly polarized light, and the demodulated light generates a second reflected light to the left through the polarization beam splitter; the second polarizer further filters the second reflected light to remove stray light and then converges it onto the target surface of the detector through the imaging lens to obtain a microscopic image of the sample.

[0017] Furthermore, through the light modulation of the turntable, the axial resolution of the microscope is improved. By moving the sample stage, multiple extremely thin optical section images can be obtained. As the objective lens focuses on a certain point on the sample surface, the light intensity at this point will increase sharply, and when out of focus, the brightness will quickly decay to a position close to 0. Therefore, each point on the sample surface has a Gaussian distribution in the axial direction.

[0018] Furthermore, the turntable has multiple partitions, and confocal composite images and wide-field images can be collected separately. When rotating the turntable to scan the sample, confocal composite images and wide-field images of the same focal plane are collected simultaneously, and image calculation can be performed to remove the influence of crosstalk between apertures, obtaining a pure confocal image with high contrast.

[0019] As an example, the confocal composite image includes a pure confocal image and a part of the wide-field image.

[0020] A three-dimensional reconstruction method based on an aperture-related confocal microscopy measurement system collects microscopic images of the sample at different height positions by moving the displacement stage along the z-axis, reconstructs the three-dimensional topography of the sample surface by combining the microscopic images at different height positions of the sample, can measure the microscopic three-dimensional features of the sample, and can also detect surface defects of the sample. The processing process includes:

[0021] S1: According to the measured sample, set the scanning spacing and fitting degree requirements of the system.

[0022] S2: Collect confocal composite images and wide-field images of the sample 112 at different heights.

[0023] S3: Create an image stack to store the information of the pure confocal images of each focal plane.

[0024] Create a height bitmap to store the height position information of each point on the sample surface.

[0025] S4: Read the confocal composite images and wide-field images of the sample 112 at each height position in sequence.

[0026] S5: Subtract the wide-field image with a certain proportional coefficient from the confocal composite image to obtain the pure confocal image at each height, and store it in the corresponding position of the image stack. The certain proportional coefficient depends on the specific turntable design parameters in the optical path system, and the selection criterion is to remove background interference.

[0027] S6: The pixels at the same position in each image in the image stack form the axial response curve of the corresponding sample point. Perform data fitting on the axial response curve, locate the peak position, and store the information of the peak position in the corresponding pixel position of the height bitmap.

[0028] As an example, the peak position refers to the actual height of the corresponding point of the sample;

[0029] Further, the method adopted for data fitting is a Gaussian fitting method based on the weighted least squares and its iterative processing method, including:

[0030] In order to achieve high-precision measurement of the sample profile, it is necessary to perform Gaussian fitting on the axial response curve of the sample; the Gaussian model function is:

[0031]

[0032] where f(z) represents the signal intensity, A represents the curve height, μ represents the position of the central axis of the curve (i.e., the maximum value), and σ is the standard deviation, which is used to represent the curve width;

[0033] After converting the nonlinear Gaussian function into a linear quadratic function in the existing least squares Gaussian fitting, the least squares linear fitting is then performed; taking the logarithm of both sides of the Gaussian function gives:

[0034]

[0035] Let lnf(z) = y; Then there is a quadratic polynomial

[0036] y = az 2 + bz + c Formula (3)

[0037] According to the least squares principle, substituting the coordinate values (z i , y i ) of each optical slice at a certain point into the equation gives the sum of squared differences as:

[0038] V 2 = ∑(f(z i ) - y i ) 2 = ∑(az i 2 + bz i + c - y i ) 2 = min Formula 4

[0039] where: z i refers to the position of the optical slice in the axial direction, and y i refers to the natural logarithm of the signal intensity of the point in this layer;

[0040] At this time, the problem is transformed into finding the extreme value of a multivariate function, and taking the derivatives of the parameters a, b, and c respectively

[0041]

[0042] It can be converted into a matrix equation

[0043]

[0044] The values of a, b, and c can be obtained by solving the system of equations, and thus

[0045]

[0046] This method converts the non - linear fitting problem into a linear fitting problem, with an extremely fast fitting speed, but relatively poor anti - noise interference ability;

[0047] In the actual usage scenario, considering the situation where there is random noise in the system, the measured light intensity value is not the true value;

[0048]

[0049] At this time, similar to Formula 4, the mean squared error of a single point can be obtained as

[0050]

[0051] So the calculable V 2 The expected value of is

[0052]

[0053] where σ Δ 2 is the standard deviation of the noise. It can be seen from Formula 9 that the smaller the value of f(z), the greater the introduced error. The weighted least - squares method re - defines the error equation as:

[0054]

[0055] Therefore

[0056] E(V 2 ) = f(z) 2 [az i 2 + bz i + c - ln f(z)] 2 + σ Δ 2 Formula 11

[0057] At this time, the noise is not affected by the value of f(z), minimizing the noise interference. Formula 6 becomes:

[0058]

[0059] Solve for the values of a, b, and c and substitute them into Formula 7 to obtain the fitting result;

[0060] S7: Calculate the goodness of fit and compare it with the pre-set threshold requirement for the goodness of fit. Under large noise conditions, there will be cases where the threshold requirement is not met. At this time, introduce its iterative processing method to perform iterative processing operations on the fitting results that do not meet the goodness of fit requirement; output the pixel height for the fitting results that meet the goodness of fit requirement.

[0061] Observed value That is the predicted value for the first time, and there is no need to predict the initial parameters like in the non-linear fitting algorithm;

[0062] For the k(k>0)-th iteration, there is

[0063]

[0064] Where: a k-1 , b k-1 , c k-1 are the calculation results of the previous iteration, and f(z) k is substituted as the new iterative predicted value into the position of Equation (12) for calculation. After each iteration, continue to compare with the threshold until the requirement is met and then output; if the goodness of fit does not change or deteriorates continuously for multiple iterations, output 0, indicating that it cannot be measured;

[0065] S8: Detect each pixel one by one to see if the pixel height output is completed. When the height output of all pixels is completed, that is, all pixels in the height bitmap are filled, output the height bitmap;

[0066] S9: Use the gray value of each pixel in the height bitmap as the z-axis height to 3Dize the height bitmap and output the three-dimensional contour map of the sample; the gray value is the height position information; The beneficial effects of the present invention:

[0067] The present invention proposes a three-dimensional reconstruction algorithm for a confocal microscope related to the aperture. Based on the Gaussian fitting of the weighted least squares method and its iterative processing method, it can quickly and accurately fit the height position of the surface points of the sample.

[0068] The system structure of the present invention is scientifically optimized, easy to implement, convenient to maintain, and suitable for popularization. Description of the Drawings

[0069] Figure 1 is the overall structure schematic diagram of the confocal microscopic measurement system related to the aperture of the present invention.

[0070] Figure 2 is the principle design reference block diagram of the three-dimensional reconstruction method of the confocal microscopic measurement system related to the aperture of the present invention. Detailed Embodiment

[0071] The preferred embodiments of the present invention will be described in detail below with reference to the drawings.

[0072] Referring to Figure 1 as shown, an aperture-related confocal microscopic measurement system and its three-dimensional reconstruction method, wherein:

[0073] The aperture-related confocal microscopic measurement system includes:

[0074] a light source 101, a first collimating lens 102, a first polarizer 103, a polarization beam splitter 104, a turntable 105, a tube lens 106, a quarter-wave plate 107, an objective lens 108, a second polarizer 109, an imaging lens 110, and a detector 111;

[0075] The first collimating lens 102 is disposed below the light source 101, the first polarizer 103 is disposed below the first collimating lens 102, the polarization beam splitter 104 is disposed below the first polarizer 103, the turntable 105 is disposed below the polarization beam splitter 104, the tube lens 106 is disposed below the turntable 105, the quarter-wave plate 107 is disposed below the tube lens 106, the objective lens 108 is disposed below the quarter-wave plate 107, and a sample 112 is placed below the objective lens 108;

[0076] As an example, the sample 112 is placed on a displacement stage.

[0077] The second polarizer 109 is disposed on the left side of the polarization beam splitter 104, the imaging lens 110 is disposed on the left side of the second polarizer 109, and the detector 111 is disposed on the left side of the imaging lens 110; and it is ensured that the optical axis of the second polarizer 109 is orthogonal to the optical axis of the first polarizer 103;

[0078] The light source 101 is used to emit illumination light. After being collimated by the collimating lens 102, the illumination light passes through the first polarizer 103 to generate incident linearly polarized light; the incident linearly polarized light is transmitted at the polarization beam splitter 104, and the transmitted light is modulated by the turntable 105. The modulated light is focused on the rear focal plane of the objective lens 108 by the tube lens 106 again. During this period, the illumination light passes through the quarter-wave plate 107 and becomes circularly polarized light; the circularly polarized light is then projected onto the sample 112 by the objective lens 108;

[0079] The reflected light generated on the sample 112 returns along the original path. At the quarter-wave plate 107, it becomes a reflected linearly polarized light whose vibration direction is orthogonal to that of the incident light. Then, it is focused on the turntable 105 through the lens barrel lens 106. The turntable 105 demodulates the reflected linearly polarized light, and the demodulated light generates a second reflected light to the left through the polarization beam splitter 104. The second polarizer 109 further filters the second reflected light to remove stray light and then converges it to the target surface of the detector 111 through the imaging lens 110 to obtain a microscopic image of the sample.

[0080] Furthermore, through the optical modulation of the turntable 105, the axial resolution of the microscope is improved. By moving the sample 112 with the displacement stage, multiple extremely thin optical section images can be obtained. As the objective lens 108 focuses on a certain point on the surface of the sample 112, the light intensity at this point will increase sharply, and when out of focus, the brightness will rapidly decay to a position close to 0. Therefore, each point on the surface of the sample 112 has a Gaussian distribution axially.

[0081] Furthermore, the turntable 105 has multiple partitions, which can respectively collect confocal composite images and wide-field images. When rotating the turntable 105 to scan the sample 112, the confocal composite image and the wide-field image are simultaneously collected for the same focal plane, and image calculation can remove the influence of crosstalk between apertures to obtain a pure confocal image with high contrast.

[0082] As an example, the confocal composite image includes a pure confocal image and a part of the wide-field image.

[0083] Refer to Figure 2 As shown, for the three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system, by moving the displacement stage along the z-axis to collect microscopic images of the sample at different height positions, and combining the microscopic images of the sample at different height positions to reconstruct the three-dimensional topography of the sample surface, the microscopic three-dimensional characteristics of the sample can be measured, and the surface defects of the sample can also be detected. The processing process includes:

[0084] S1: According to the measured sample, set the scanning pitch and fitting degree requirements of the system.

[0085] S2: Collect the confocal composite images and wide-field images of the sample 112 at different heights.

[0086] S3: Create an image stack to store the pure confocal image information of each focal plane.

[0087] Create a height bitmap to store the height position information of each point on the sample surface.

[0088] S4: Sequentially read the confocal composite images and wide-field images of the sample 112 at each height position.

[0089] S5: Subtract the wide-field image with a certain proportional coefficient from the confocal composite image to obtain the pure confocal image at each height, and store it in the corresponding position of the image stack; the certain proportional coefficient depends on the specific design parameters of the spinning disk in the optical path system, and the selection criterion is to remove background interference;

[0090] As an example, when the stripe on the spinning disk has a light-transmitting aperture of 40 μm and an aperture interval of 120 μm, the value of the certain proportional coefficient is: 0.3.

[0091] S6: The pixels at the same position of each image in the image stack form the axial response curve of the corresponding sample point. Perform data fitting on the axial response curve, locate the peak position, and store the information of the peak position in the corresponding pixel position of the height bitmap;

[0092] As an example, the peak position refers to the actual height of the corresponding point of the sample;

[0093] Further, the method used for data fitting is the Gaussian fitting method based on the weighted least squares and its iterative processing method, including:

[0094] In order to achieve high-precision measurement of the sample profile, it is necessary to perform Gaussian fitting on the axial response curve of the sample; the Gaussian model function is:

[0095]

[0096] where f(z) represents the signal intensity, A represents the curve height, μ represents the position of the central axis of the curve (i.e., the maximum value), and σ is the standard deviation, which is used to represent the curve width;

[0097] The existing least squares Gaussian fitting converts the non-linear Gaussian function into a linear quadratic function and then performs least squares linear fitting; taking the logarithm of both sides of the Gaussian function gives:

[0098]

[0099] Let lnf(z) = y; Then there is a quadratic polynomial

[0100] y = az 2 + bz + c Formula (3)

[0101] According to the least squares principle, substituting the coordinate values (z i , y i ) of each optical section at a certain point into the equation gives the square difference as:

[0102] V 2 = ∑(f(z i ) - y i )2 = ∑(az i 2 + bz i + c - y i ) 2 = min Formula 4

[0103] Where: z i refers to the position of the optical slice in the axial direction, and y i refers to the natural logarithm of the signal intensity of the point in this layer;

[0104] At this time, the problem is transformed into finding the extreme value of a multivariate function, and the partial derivatives of the parameters a, b, and c are calculated respectively

[0105]

[0106] It can be converted into a matrix equation

[0107]

[0108] Solving the system of equations can obtain the values of a, b, and c, from which we can get

[0109]

[0110] This method transforms the non - linear fitting problem into a linear fitting problem, with an extremely fast fitting speed, but relatively poor anti - noise interference ability;

[0111] In the actual usage scenario, considering the situation where there is random noise in the system, the measured light intensity value is not the true value;

[0112]

[0113] At this time, similar to Formula 4, the mean square error of a single point can be obtained as:

[0114]

[0115] So the expected value of the computable V 2 is:

[0116]

[0117] Where, σ Δ 2 is the standard deviation of the noise. It can be seen from Formula 9 that the smaller the value of f(z), the greater the introduced error. The weighted least - squares method re - defines the error equation as:

[0118]

[0119] Therefore

[0120] E(V 2) = f(z) 2 [az i 2 +bz i 2 +c - ln f(z)] 2 +σ Δ 2 Formula 11

[0121] At this time, the noise is not affected by the value of f(z), minimizing the noise interference. Formula 6 becomes:

[0122]

[0123] Solve for the values of a, b, and c and substitute them into Formula 7 to obtain the fitting result;

[0124] S7: Calculate the goodness of fit and compare it with the preset goodness-of-fit requirement threshold. Under large noise conditions, there may be cases where the threshold requirement is not met. At this time, introduce its iterative processing method to perform iterative processing operations on the fitting results that do not meet the goodness-of-fit requirements; output the pixel height for the fitting results that meet the goodness-of-fit requirements.

[0125] Observed value Is the predicted value for the first time and does not require predicting the initial parameters like the non - linear fitting algorithm;

[0126] For the k - th (k > 0) iteration, there is

[0127]

[0128] Where: a k-1 、b k-1 、c k-1 Are the calculation results of the previous iteration, and f(z) k Is substituted as the new iterative predicted value into the position in Equation (12) for calculation. After each iteration, continue to compare with the threshold until the requirement is met and then output; if the goodness of fit does not change or deteriorates after consecutive multiple iterations, output 0, indicating that it cannot be measured;

[0129] S8: Detect each pixel one by one to see if the pixel height output is completed. When the height output of all pixels is completed, that is, all pixels in the height bitmap are filled, output the height bitmap;

[0130] S9: Use the gray - scale value of each pixel in the height bitmap as the z - axis height to 3Dize the height bitmap and output the three - dimensional contour map of the sample; the gray - scale value is the height position information;

[0131] ​The present invention proposes a three-dimensional reconstruction algorithm for a confocal microscope related to aperture diameters. Based on Gaussian fitting of the weighted least squares method and its iterative processing method, it can quickly and highly accurately fit the height positions of the surface points of the sample. The system structure of the present invention is scientifically optimized, easy to implement, convenient to maintain, and suitable for popularization.

[0132] The above are only the preferred embodiments of the present invention. It should be understood that the description of the above embodiments is only used to help understand the method and its core idea of the present invention, and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. 3D reconstruction method based on an aperture-correlation confocal microscopy measurement system, characterized in that, it includes: an aperture-correlation confocal microscopy measurement system, which is provided with: a light source, a first collimating lens, a first polarizer, a polarization beam splitter, a turntable, a tube lens, a quarter-wave plate, an objective lens, a second polarizer, an imaging lens, and a detector; the first collimating lens is arranged below the light source, the first polarizer is arranged below the first collimating lens, the polarization beam splitter is arranged below the first polarizer, the turntable is arranged below the polarization beam splitter, the tube lens is arranged below the turntable, the quarter-wave plate is arranged below the tube lens, the objective lens is arranged below the quarter-wave plate, and a sample is placed below the objective lens; the sample is placed on a displacement stage; the second polarizer is arranged on the left side of the polarization beam splitter, the imaging lens is arranged on the left side of the second polarizer, and the detector is arranged on the left side of the imaging lens; and it is ensured that the optical axis of the second polarizer is orthogonal to the optical axis of the first polarizer; by moving the displacement stage along the z-axis, microscopic images of the sample at various height positions are collected, and the three-dimensional topography of the sample surface is reconstructed by combining the microscopic images of the sample at various height positions. The microscopic three-dimensional features of the sample can be measured, and the surface defects of the sample can also be detected. The processing process includes: S1: According to the measured sample, set the scanning pitch and fitting degree requirements of the system; S2: Collect the confocal composite images and wide-field images of the sample at various heights; S3: Create an image stack for storing the pure confocal image information of each focal plane; Create a height bitmap for storing the height position information of each point on the sample surface; S4: Sequentially read the confocal composite images and wide-field images of the sample at each height position S5: Subtract the wide-field image with a certain proportional coefficient from the confocal composite image to obtain the pure confocal image at each height, and store it in the corresponding position of the image stack; the certain proportional coefficient depends on the specific turntable design parameters in the optical path system, and the selection criterion is to remove background interference; S6: The pixels at the same position in each image in the image stack form the axial response curve of the corresponding sample point. Perform data fitting on the axial response curve, locate its peak position, and store the information of the peak position in the corresponding pixel position of the height bitmap; the peak position refers to the actual height of the corresponding point of the sample; the method used for data fitting is the Gaussian fitting method based on the weighted least squares and its iterative processing method; S7: Calculate the fitting degree and compare it with the preset fitting degree requirement threshold. Under large noise conditions, there will be cases where the threshold requirement is not met. At this time, introduce its iterative processing method to perform operations on the fitting results that do not meet the fitting degree requirement; output the pixel height for the fitting results that meet the fitting degree requirement; S8: Detect each pixel one by one to see if the pixel height output is completed. When the height output of all pixels is completed, that is, when all pixels in the height bitmap are filled, output the height bitmap; S9: Use the grayscale value of each pixel in the height bitmap as the z-axis height, 3Dize the height bitmap, and output the three-dimensional contour map of the sample; the grayscale value is the height position information.

2. The three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system according to claim 1, wherein, the light source is used to emit illumination light. After being collimated by the collimating lens, the illumination light passes through the first polarizer to generate incident linearly polarized light; the incident linearly polarized light is transmitted at the polarization beam splitter, and the transmitted light is modulated by the turntable. The modulated light is focused on the rear focal plane of the objective lens by the tube lens again. During this period, the illumination light passes through the quarter-wave plate and becomes circularly polarized light; the circularly polarized light is projected onto the sample by the objective lens; The reflected light generated on the sample returns along the original path. At the quarter-wave plate, it becomes reflected linearly polarized light orthogonal to the vibration direction of the incident light, and is focused on the turntable by the tube lens. The turntable demodulates the reflected linearly polarized light, and the demodulated light generates a second reflected light to the left through the polarization beam splitter; the second polarizer further filters the second reflected light to remove stray light and then converges on the target surface of the detector through the imaging lens to obtain the microscopic image of the sample.

3. The three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system according to claim 2, wherein, the turntable has multiple partitions for separately collecting confocal composite images and wide-field images.

4. The three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system according to claim 3, wherein, When rotating the turntable to scan the sample, the confocal composite image and the wide-field image are collected simultaneously on the same focal plane, and image calculation is performed to remove the influence of crosstalk between apertures, and a high-contrast pure confocal image is obtained.

5. The aperture-related confocal microscopy measurement system according to claim 4, wherein, the confocal composite image includes a pure confocal image and a part of the wide-field image.

6. The three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system according to claim 1, wherein, the method adopted for data fitting includes: In order to achieve high-precision measurement of the sample contour, it is necessary to perform Gaussian fitting on the axial response curve of the sample; the Gaussian model function is: where f(z) represents the signal intensity, A represents the curve height, μ represents the position of the central axis of the curve, and σ is the standard deviation, which is used to represent the curve width; The existing least squares Gaussian fitting converts the nonlinear Gaussian function into a linear quadratic function and then performs least squares linear fitting; taking the logarithm of both sides of the Gaussian function gives: Let $\ln f(z)=y$; Then there is a quadratic polynomial: y = az 2 + bz + c Equation 3 According to the least squares principle, substituting the coordinate values (z i , y i ) of a certain point on each optical section into the equation gives the sum of squared differences as: V 2 = ∑(f(z i ) - y i ) 2 = ∑(az i 2 + bz i + c - y i ) 2 = min Formula 4 where: z i represents the position of the optical slice in the axial direction, and y i represents the natural logarithm of the signal intensity of the point in this layer; At this time, the problem is transformed into finding the extreme value of a multivariate function, and the partial derivatives of the parameters a, b, and c are calculated respectively: Convert it into a matrix equation: Solve the system of equations to obtain the values of a, b, and c, and thus obtain: This method converts the nonlinear fitting problem into a linear fitting problem, has an extremely fast fitting speed, but has relatively poor anti-noise interference ability; In the actual use scenario, considering the situation where there is random noise in the system, the measured light intensity value is not the true value: At this time, the mean square error of a single point obtained based on formula 4 is: The calculable V at this time 2 has an expected value of: where, σ Δ 2 is the standard deviation of the noise; it can be seen from Equation 9 that the smaller the value of f(z), the greater the introduced error; the weighted least squares method redefines the error equation as: Therefore E(V 2 ) = f(z) 2 [az i 2 + bz i + c - ln f(z)] 2 + σ Δ 2 Equation 11 At this time, the noise is not affected by the value of f(z), minimizing the noise interference; Equation 6 becomes: Solve for the values of a, b, and c and substitute them into Equation 7 to obtain the fitting result.

7. The three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system according to claim 6, characterized in that the iterative processing method includes: Observed value That is the predicted value for the first time and there is no need to predict the initial parameters like the non-linear fitting algorithm; For the k-th iteration, where k > 0, there is: where: a k-1 , b k-1 , c k-1 are the calculation results of the previous iteration, and f(z) k is substituted as the new iteration prediction value into the position in Formula 12 for calculation; after each iteration, continue to compare with the threshold until the requirement is met and then output; if the fitting degree does not change or deteriorates continuously for multiple iterations, output a value of 0, indicating that it cannot be measured.

8. The three-dimensional reconstruction method based on the aperture-related confocal microscopy measurement system according to claim 7, characterized in that when the stripe on the turntable has a light-transmitting aperture of 40 μm and an aperture interval of 120 μm, the value of the certain proportionality coefficient is: 0.3.

Images