A method for eliminating spot aliasing in line spectrum confocal images based on Wiener filtering
By constructing an improved Wiener filter and point diffusion function, combined with weighted fusion splicing, the spot aliasing problem of linear spectral confocal sensors is solved, the measurement accuracy and resolution are improved, the algorithm is simplified, and it is suitable for a variety of imaging scenarios.
Patent Information
- Application Number
- CN202311664233.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-06
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2043-12-06
AI Technical Summary
Linear spectral confocal sensors have spot aliasing phenomenon during imaging, which affects measurement accuracy. The prior art such as blind deconvolution methods are complex and have limited effects.
Using a method based on Wiener filter, an improved Wiener filter is constructed, and the point diffusion function is used to perform approximate estimation and inverse Fourier transform in the frequency domain, combined with weighted fusion splicing, eliminate spot aliasing and improve measurement resolution.
The algorithm is simplified, and the horizontal measurement accuracy of the linear spectral confocal sensor is improved, and the effect is better. It is suitable for difficult-to-estimate imaging scenes and meets the real usage conditions.
Smart Images

Figure CN117808704B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of precision optical measurement, and in particular to a method for eliminating light spot aliasing of a line spectrum confocal image based on Wiener filtering. Background Art
[0002] Line spectrum confocal sensor is an optical non-contact displacement sensor that has emerged in the field of precision optical measurement in recent years. It has the characteristics of being able to break through the optical diffraction limit to achieve super-resolution measurement, and is widely used in the field of industrial automation. The working principle of the line spectrum confocal sensor is to focus the ideal line light source after passing through the slit at different depths with different wavelengths, image the light reflected from the sample surface into the imaging spectrometer for spectroscopic decoding and obtain spectral image data, and finally restore the surface morphology data of the measured sample based on the spectral image data. However, due to the characteristics of the line spectrum confocal sensor itself, the spectral image usually has spot aliasing, that is, during the imaging process, the actual spot is larger than the ideal spot, and adjacent spots will partially overlap. This spot aliasing phenomenon will affect the measurement accuracy of the line spectrum confocal sensor, so how to eliminate spot aliasing is a very important topic.
[0003] Existing line spectrum confocal sensors generally directly use the data processing method of spectrum peak extraction. This processing method is to find the peak point of the spectrum data on each spatial channel of the spectrum image, and then infer the surface morphology data of the sample through the peak point position obtained based on the model and calibration results. However, due to the existence of spot aliasing, the extracted spectrum peak position may not be the actual position, which will greatly affect the measurement accuracy.
[0004] Existing methods for removing blur from images attempt to achieve this through blind deconvolution, which is a general image restoration method. This method attempts to analyze the spectrum of the blurred image, determine the transfer function of the signal environment, and then restore the original data through deconvolution to remove image blur. For example, the invention patent with publication number CN116309190B and name "A turbulence-degraded image restoration method based on the median prior of the optimal region" first obtains the target point spread function based on the gradient information of the target turbulence-degraded image and the blind deconvolution model, and then performs blind deconvolution processing on the target turbulence-degraded image to achieve image restoration. However, this processing method is not only complex, but also a high-intensity computing technology, and it only relies on the analysis of the spectrum of the image data, and its final effect still needs to be improved. Summary of the invention
[0005] In view of the deficiencies in the prior art, the present invention proposes a method for eliminating light spot aliasing of a line spectrum confocal image based on Wiener filtering.
[0006] The present invention is implemented by the following technical solution, which includes five steps, which are specifically described as follows:
[0007] Step 1: Use the line light source of the line spectrum sensor to illuminate the surface of the sample to be tested, and collect the reflected light through the spectrometer spectrometer system of the line spectrum sensor, and focus it on the CMOS sensor after spectroscopic processing;
[0008] Step 2: Read the data on the CMOS sensor to obtain the original spectral image data I(x,y) of the line spectrum confocal sensor presented in pixel points, where x and y are pixel coordinates.
[0009] Step 3: Replace the sample to be tested with a probe with a small cross-sectional area. The radius of a spot on the probe and the sample to be tested differs by 0-7.5μm. The probe is fixed on a precision displacement platform. By adjusting the horizontal position and height of the probe, N*M evenly distributed non-aliased spots are obtained on the CMOS sensor, corresponding to M longitudinal probe heights at each of the N transverse probe positions.
[0010] Step 4: For N*M non-aliased light spots, read N*M groups of CMOS pixel data near them to obtain the measured point spread function H of each light spot i (x,y);
[0011] Step 5: Divide the CMOS sensor plane into P*Q regions, and calculate the point spread function H(x, y) based on the pixel coordinates measured in each region;
[0012] Step 6: Based on the original spectral image data described in step 2 and the P*Q point spread functions obtained by actual measurement, each of the divided P*Q regions is first expanded to form P*Q regions with overlapping L columns and rows, and improved Wiener filters are respectively constructed for these regions, and Wiener filtering is performed to obtain P*Q line spectrum confocal sensor spectral images after spot aliasing is eliminated;
[0013] The improved Wiener filter is constructed to perform Wiener filtering, and the specific process includes:
[0014] Step 6-1: Construct a Wiener filter and make an approximate estimate in the frequency domain. The expression formula is as follows:
[0015]
[0016] Among them, H(u,v) is the spectrum of the point spread function, H * (u,v) is its complex conjugate, is a certain empirical value, I(u,v) is the spectrum of the noisy spectral data, is an approximate estimate of the image spectrum. Performing an inverse Fourier transform, we can get an estimate of the image in the time domain
[0017] Step: 6-2: Construct a modified Wiener filter, the expression formula is as follows:
[0018]
[0019] Among them, H * (u,v) is the complex conjugate of the spectrum function of the point spread function, H(u,v) is the spectrum of the point spread function, I(u,v) is the spectrum of the noisy spectrum data, O(u,v) is the spectrum of the original spectrum data, is the mean square value of noise, S o (u,v) is the estimated value from step 6-1 The power spectral density spectrum. The estimation method is expressed as:
[0020]
[0021] Where M and N are the width and height of the image, respectively, i(x,y) is the spectral data with noise, is an estimate of the original image obtained from step 6-1, and h(x,y) is the point spread function. Perform an inverse Fourier transform on O(u,v) to obtain the line spectrum confocal sensor spectral image with spot aliasing eliminated, thereby effectively improving the horizontal and vertical measurement resolution of the line spectrum confocal sensor.
[0022] Step 7: Perform weighted fusion and splicing on the overlapping areas of the P*Q processed spectral images to eliminate the influence of the filtering boundary.
[0023] Furthermore, in step 3, by adjusting the horizontal position and height of the light spot, N*M evenly distributed light spots without overlapping each other are obtained on the CMOS sensor. The specific content is:
[0024] After the probe under test is measured by the line spectrum sensor, it is imaged as a non-aliasing light spot on the CMOS sensor. By adjusting the position and height of the probe, a total of N*M light spots are taken at different positions on the CMOS sensor, where N represents N positions equidistantly taken within the lateral measurement range of the line spectrum sensor used, and M represents M heights equidistantly taken within the longitudinal measurement range of the line spectrum sensor used.
[0025] Furthermore, the CMOS sensor plane is divided into P*Q regions as described in step 5, and the point spread function H(x, y) based on the pixel coordinates measured in each region is calculated. The specific process includes:
[0026] Step 5-1: Considering that the light spot diffusion functions at different positions on the CMOS sensor are different, the CMOS sensor plane is evenly divided into P*Q regions, where P means that the CMOS sensor plane is evenly divided into P regions in the horizontal direction, and Q means that the CMOS sensor plane is evenly divided into Q regions in the vertical direction;
[0027] Step 5-2: In each divided area, for the R light spots falling in the area, according to the point spread function H described in step 4 i (n,m), sum and average the corresponding pixels to obtain the point spread function H(x,y) based on the pixel coordinates in the area, that is,
[0028] Furthermore, the weighted fusion splicing described in step 7 specifically includes the following process:
[0029] For a pixel point in the overlapping area of adjacent images, its value is calculated as follows:
[0030]
[0031] Among them, P1 is the value of the pixel in the left or upper image of the adjacent image, P2 is the value of the pixel in the right or lower image of the adjacent image, P is the value after the overlapping area is fused, l represents the distance between a certain pixel in the overlapping area and the rightmost or lowermost boundary line of the overlapping area, and L represents the width of the overlapping area and depends on the size of the light spot. Since the pixels in the overlapping area are affected by two point spread functions at the same time, this method can achieve better restoration of the pixels in the overlapping area.
[0032] Preferably, the probe with a small cross-sectional area has a radius of 5-20 μm.
[0033] The beneficial effects of the present invention are:
[0034] (1) The spot aliasing phenomenon of the line spectrum sensor is much more serious than that of the point spectrum sensor. The image after eliminating the spot aliasing is closer to the real data than the original spectral data image, which helps to improve the measurement accuracy of the line spectrum confocal sensor in both the lateral and axial dimensions.
[0035] (2) Compared with the blind deconvolution method based on spectrum analysis, the method based on point spread function and Wiener filtering not only does not require a large amount of analysis and calculation of the original spectral image data, simplifies the algorithm, but also has a better effect in eliminating spot aliasing and more accurately restores the ideal spectral image.
[0036] (3) Compared with the blind deconvolution method based on spectrum analysis, the method based on point spread function and Wiener filtering can still effectively eliminate the spot aliasing phenomenon in imaging scenarios that are difficult to estimate, such as when the samples are sparse or have obvious periodicity, because it does not rely on spectrum analysis.
[0037] (4) The method of obtaining the point spread function based on actual measurement is more in line with the actual line spectrum confocal sensor system, so it has a good spot aliasing elimination effect. The method of sampling at different imaging positions takes into account the slight difference in aberrations at different positions. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 It is a flow chart of the method for eliminating light spot aliasing of line spectrum confocal image based on Wiener filtering of the present invention;
[0039] Figure 2 It is a schematic diagram of the point spread function of the present invention;
[0040] Figure 3 This is an example diagram of the light spot aliasing phenomenon generated in an example of the present invention;
[0041] Figure 4 This is a diagram showing the effect of eliminating light spot aliasing based on point spread function and Wiener filtering in an example of the present invention. DETAILED DESCRIPTION
[0042] In order to make the purpose, technical method and advantages of the present invention clearer, the specific implementation of the present invention is further described in detail below with reference to the accompanying drawings and examples. However, it should be understood that the embodiments specifically described herein are only used to explain the present invention and are not intended to limit the scope of protection of the present invention in any way.
[0043] The present invention discloses a method for eliminating light spot aliasing of a line spectrum confocal image based on Wiener filtering, such as Figure 1 As shown in the flowchart, it includes the following seven steps:
[0044] Step 1: Use the line light source of the line spectrum sensor to illuminate the surface of the sample to be tested, and collect the reflected light through the spectrometer spectrometer system of the line spectrum sensor, and focus it on the CMOS sensor after spectroscopic processing;
[0045] Step 2: Read the data on the CMOS sensor to obtain the original spectral image data I(x,y) of the line spectrum confocal sensor presented in pixel points, where x and y are pixel coordinates.
[0046] Step 3: Replace the sample to be tested with a probe with a small cross-sectional area, the cross-sectional area is similar to the size of a light spot on the sample to be tested, the probe is fixed on a precision displacement platform, and N*M evenly distributed non-aliased light spots are obtained on the CMOS sensor by adjusting the horizontal position and height of the probe, corresponding to M longitudinal probe heights at each of the N transverse probe positions; the specific content of the adjustment of the horizontal position and height of the light spot to obtain N*M evenly distributed non-aliased light spots on the CMOS sensor is as follows:
[0047] After the probe under test is measured by the line spectrum sensor, it is imaged as a non-aliasing light spot on the CMOS sensor. By adjusting the position and height of the probe, a total of N*M light spots are taken at different positions on the CMOS sensor, where N represents N positions equidistantly taken within the lateral measurement range of the line spectrum sensor used, and M represents M heights equidistantly taken within the longitudinal measurement range of the line spectrum sensor used.
[0048] Step 4: For N*M non-aliased light spots, read N*M groups of CMOS pixel data near them to obtain the measured point spread function H of each light spot i (x,y);
[0049] Step 5: Divide the CMOS sensor plane into P*Q regions, and calculate the point spread function H(x,y) based on the pixel coordinates measured in each region; the specific process includes:
[0050] Step 5-1: Considering that the light spot diffusion functions at different positions on the CMOS sensor are different, the CMOS sensor plane is evenly divided into P*Q regions, where P means that the CMOS sensor plane is evenly divided into P regions in the horizontal direction, and Q means that the CMOS sensor plane is evenly divided into Q regions in the vertical direction;
[0051] Step 5-2: In each divided area, for the R light spots falling in the area, according to the point spread function H described in step 4 i (n,m), sum and average the corresponding pixels to obtain the point spread function H(x,y) based on the pixel coordinates in the area, that is,
[0052] Step: 6: Based on the original spectral image data described in step 2 and the P*Q point spread functions obtained by actual measurement, each of the divided P*Q areas is first expanded to form P*Q areas with overlapping L columns and rows, and improved Wiener filters are constructed for these areas respectively, and Wiener filtering is performed to obtain P*Q line spectrum confocal sensor spectral images after eliminating spot aliasing.
[0053] The improved Wiener filter is constructed to perform Wiener filtering, and the specific process includes:
[0054] Step 6-1: Construct a Wiener filter and make an approximate estimate in the frequency domain. The expression formula is as follows:
[0055]
[0056] Among them, H(u,v) is the spectrum of the point spread function, H * (u,v) is its complex conjugate, is a certain empirical value, I(u,v) is the spectrum of the noisy spectral data, is an approximate estimate of the frequency spectrum of the image. Performing an inverse Fourier transform, we can get an estimate of the image in the time domain
[0057] Step: 6-2: Construct a modified Wiener filter, the expression formula is as follows:
[0058]
[0059] Among them, H * (u,v) is the complex conjugate of the spectrum function of the point spread function, H(u,v) is the spectrum of the point spread function, I(u,v) is the spectrum of the noisy spectrum data, O(u,v) is the spectrum of the original spectrum data, is the mean square value of noise, S o (u,v) is the estimated value from step 6-1 The power spectral density spectrum. The estimation method is expressed as:
[0060]
[0061] Where M and N are the width and height of the image, respectively, i(x,y) is the spectral data with noise, is an estimate of the original image obtained from step 6-1, and h(x,y) is the point spread function. Perform an inverse Fourier transform on O(u,v) to obtain the line spectrum confocal sensor spectral image with spot aliasing eliminated, thereby effectively improving the horizontal and vertical measurement resolution of the line spectrum confocal sensor.
[0062] Step 7: Perform weighted fusion and splicing on the overlapping areas of the P*Q processed spectral images to eliminate the influence of the filtering boundary.
[0063] The weighted fusion splicing process includes:
[0064] For a pixel point in the overlapping area of adjacent images, its value is calculated as follows:
[0065]
[0066] Among them, P1 is the value of the pixel in the left or upper image of the adjacent image, P2 is the value of the pixel in the right or lower image of the adjacent image, P is the value after the overlapping area is fused, l represents the distance between a certain pixel in the overlapping area and the rightmost or lowermost boundary line of the overlapping area, and L represents the width of the overlapping area and depends on the size of the light spot. Since the pixels in the overlapping area are affected by two point spread functions at the same time, this method can achieve better restoration of the pixels in the overlapping area.
[0067] The point spread function is further explained below.
[0068] The physical meaning of the point spread function refers to the intensity distribution formed on the imaging plane after the light emitted from a point light source passes through a series of optical elements.
[0069] like Figure 2 As shown in the schematic diagram, the left picture is an imaging point and the spectral image nearby read by CMOS under ideal imaging conditions, and all light falls within the target pixel; the right picture is the spectral image under actual conditions. Since the light spot is diffused, the pixels near the ideal imaging point can also read the spectral data, and the point spread function is a function that describes the light intensity distribution near the ideal imaging point on the CMOS.
[0070] The following further describes the Wiener filtering method based on the original spectral image data and the point spread function.
[0071] For the spectral data affected by aliasing, it is equivalent to convolving the original spectral data with the point spread function. Therefore, the obtained spectral data I(x, y) and the point spread function H(x, y) are Fourier transformed to obtain the spectral spectrum I(u, v) and the point spread function spectrum H(u, v) in the frequency domain. Then, the expression formula of the spectral data in the frequency domain is as follows:
[0072] I(u,v)=O(u,v)H(u,v)+N(u,v)
[0073] Where I(u,v) is the spectrum of the noisy spectral data, O(u,v) is the spectrum of the original spectral data, H(u,v) is the spectrum of the point spread function, and N(u,v) is the spectrum of the noise.
[0074] At this point, a Wiener filter can be constructed to perform the first Wiener filtering. The Wiener filter expression formula is as follows:
[0075]
[0076] Among them, H w (u,v) is the Wiener filter, H * (u, v) is the complex conjugate of the spectrum function of the point spread function, H(u, v) is the spectrum of the point spread function, S oo (u,v) is the power spectrum density of the original image, S nn (u,v) is the noise power spectral density. Since the power spectral density of the original image and noise is generally difficult to obtain, usually Set to a certain experience value Experience Points The selection can be done by selecting different Values, in these different The image is filtered at different values to obtain different restoration results, and the restoration effect is observed. Selection.
[0077] According to the obtained Wiener filter, an approximate estimate is made in the frequency domain, and the expression formula is as follows:
[0078]
[0079] Among them, H * (u,v) is the complex conjugate of the spectrum function of the point spread function, H(u,v) is the spectrum of the point spread function, is the set empirical value, and I(u,v) is the spectrum of the noisy spectral data.
[0080] right Performing an inverse Fourier transform, we can get an estimate of the original image Then, a modified Wiener filter can be constructed, which is expressed as follows:
[0081]
[0082] Among them, H * (u,v) is the complex conjugate of the spectrum function of the point spread function, H(u,v) is the spectrum of the point spread function, is the mean square value of noise, S o (u,v) is the approximate estimate The power spectral density spectrum. The estimation method is expressed as:
[0083]
[0084] Where M and N are the width and height of the image, respectively, i(x,y) is the spectral data with noise, is an estimate of the original image obtained in , and h(x,y) is the point spread function.
[0085] According to the obtained improved Wiener filter, secondary Wiener filtering is performed to restore the spectral data. The expression formula is:
[0086] O(u,v)=H MW (u,v)×I(u,v)
[0087] Among them, O(u,v) is the spectrum of the original spectral data, H MW (u,v) is the improved Wiener filter, I(u,v) is the spectrum of the noisy spectral data. Perform inverse Fourier transform on O(u,v) to obtain the result of Wiener filtering restoration image.
[0088] The following further explains the spot aliasing phenomenon and the spot aliasing elimination effect.
[0089] In one embodiment, Figure 3 As shown in the figure, the sample surface is measured by a line spectrum confocal sensor, and the reflected light is collected by the signal acquisition system and the spectrum image is obtained after the spectroscopic processing. The figure shows a part of the spectrum image. The original spectrum image 1 in the figure is a spectrum image with spot aliasing. Compared with the spectrum image 2 without spot aliasing, there is a deviation between the spectrum peak extraction position 3 in the middle column and the actual spectrum peak position 4, which affects the measurement accuracy of the line spectrum confocal sensing measurement system. Therefore, it is necessary to eliminate the spot aliasing phenomenon in the spectrum image.
[0090] In one embodiment, the spot aliasing elimination method based on point spread function and Wiener filtering of the present invention is used to process a spectral image obtained by measuring the surface of a sample with a line spectrum sensor. The effect is as follows: Figure 4 Compare Figure 4 It can be seen from the seventh and eighth columns of data of the ideal spectral image and the measured spectral image that the peak position of the measured spectral data is shifted due to the influence of the point spread function, which is not conducive to the accuracy of the sensor. Therefore, filtering is required to make the spectral image as close to the ideal as possible. The spot aliasing elimination method of the present invention can effectively eliminate the spot aliasing phenomenon in the spectral image of the line spectrum sensor and restore the spectral peak positioning to the actual imaging position, and its effect is very close to Figure 4 The ideal spectral image in the image helps to improve the measurement accuracy of the overall system, enabling the system to measure the surface morphology of the sample more accurately.
[0091] Matters not covered by the present invention are known technologies.
[0092] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for eliminating spot aliasing in line spectrum confocal images based on Wiener filtering, characterized in that: The specific steps include: Step 1: Use the line light source of the line spectrum sensor to illuminate the surface of the sample to be tested, and collect the reflected light through the spectrometer spectrometer system of the line spectrum sensor, and focus it on the CMOS sensor after spectroscopic processing; Step 2: Read the data on the CMOS sensor to obtain the original spectral image data I(x,y) of the line spectrum confocal sensor presented in pixel points, where x and y are pixel coordinates; Step 3: Replace the sample to be tested with a probe with a small cross-sectional area. The radius of a spot on the probe and the sample to be tested differs by 0-7.5μm. The probe is fixed on a precision displacement platform. By adjusting the horizontal position and height of the probe, N*M evenly distributed non-aliased spots are obtained on the CMOS sensor, corresponding to M longitudinal probe heights at each of the N transverse probe positions. Step 4: For N*M non-aliased light spots, read N*M groups of CMOS pixel data near them to obtain the measured point spread function H of each light spot i (x,y); Step 5: Divide the CMOS sensor plane into P*Q regions, and calculate the point spread function H(x, y) based on the pixel coordinates measured in each region; Step 6: Based on the original spectral image data described in step 2 and the P*Q point spread functions obtained by actual measurement, each of the divided P*Q regions is first expanded to form P*Q regions with overlapping L columns and rows, and improved Wiener filters are respectively constructed for these regions, and Wiener filtering is performed to obtain a line spectrum confocal sensor spectral image after spot aliasing is eliminated; The improved Wiener filter is constructed to perform Wiener filtering, and the specific process includes: Step 6-1: Construct a Wiener filter and make an approximate estimate in the frequency domain. The expression formula is as follows: Among them, H(u,v) is the spectrum of the point spread function, H * (u,v) is its complex conjugate, is a certain empirical value, I(u,v) is the spectrum of the noisy spectral data, To approximate the frequency spectrum of the image; Perform an inverse Fourier transform to obtain an estimate of the image in the time domain Step: 6-2: Construct an improved Wiener filter, the expression formula is as follows: Among them, H * (u,v) is the complex conjugate of the spectrum function of the point spread function, H(u,v) is the spectrum of the point spread function, I(u,v) is the spectrum of the noisy spectrum data, O(u,v) is the spectrum of the original spectrum data, is the mean square value of noise, S o (u,v) is the estimated value from step 6-1 The power spectral density spectrum of The estimation method is expressed as: Where M and N are the width and height of the image, respectively, i(x,y) is the spectral data with noise, is an estimate of the original image obtained from step 6-1, and h(x, y) is the point spread function; O(u, v) is subjected to an inverse Fourier transform to obtain a line spectrum confocal sensor spectral image that eliminates spot aliasing, thereby effectively improving the horizontal and vertical measurement resolution of the line spectrum confocal sensor; Step 7: Perform weighted fusion and splicing on the overlapping areas of the P*Q processed spectral images to eliminate the influence of the filtering boundary.
2. The method for eliminating spot aliasing of a line spectrum confocal image based on Wiener filtering according to claim 1, characterized in that: In step 3, by adjusting the horizontal position and height of the probe, N*M evenly distributed light spots without overlapping each other are obtained on the CMOS sensor, specifically: After the probe under test is measured by the line spectrum sensor, it is imaged as a non-aliasing light spot on the CMOS sensor. By adjusting the position and height of the probe, a total of N*M light spots are taken at different positions on the CMOS sensor, where N represents N positions equidistantly taken within the lateral measurement range of the line spectrum sensor used, and M represents M heights equidistantly taken within the longitudinal measurement range of the line spectrum sensor used.
3. The method for eliminating spot aliasing of a line spectrum confocal image based on Wiener filtering according to claim 1, characterized in that: Step 5 divides the CMOS sensor plane into P*Q regions, and calculates the point spread function H(x, y) based on the pixel coordinates measured in each region. The specific process includes: Step 5-1: Considering that the light spot diffusion functions at different positions on the CMOS sensor are different, the CMOS sensor plane is evenly divided into P*Q regions, where P means that the CMOS sensor plane is evenly divided into P regions in the horizontal direction, and Q means that the CMOS sensor plane is evenly divided into Q regions in the vertical direction; Step 5-2: In each divided area, for the R light spots falling in the area, according to the point spread function H described in step 4 i (n,m), sum and average the corresponding pixels to obtain the point spread function H(x,y) based on the pixel coordinates in the area, that is, 4. The method for eliminating spot aliasing of a line spectrum confocal image based on Wiener filtering according to claim 1, characterized in that: The weighted fusion splicing described in step 7 specifically includes the following process: For a pixel point in the overlapping area of adjacent images, its value is calculated as follows: Among them, P1 is the value of the pixel point in the left image or the upper image of the adjacent image, P2 is the value of the pixel point in the right image or the lower image of the adjacent image, P is the value after the overlapping area is fused, l represents the distance between a pixel point in the overlapping area and the rightmost or lowermost boundary line of the overlapping area, and L represents the width of the overlapping area and depends on the size of the light spot.
5. The method for eliminating spot aliasing of a line spectrum confocal image based on Wiener filtering according to claim 1, characterized in that: The probe with a small cross-sectional area has a radius of 5-20 μm.
