Four-wave lateral shear interference wavefront restoration method based on rhombus Gaussian window function

By using rhombus Gaussian window function filtering processing in the four-wave transverse shear interference method, noise and rank spectrum crosstalk are suppressed, and the problems of low signal-to-noise ratio and spectrum leakage are solved, achieving higher phase recovery accuracy and smoother Phase reconstruction results.

CN119085481APending Publication Date: 2024-12-06HUBEI UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310663109.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-06-05
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing four-wave transverse shear interference method is susceptible to noise and other stage spectrum crosstalk during phase extraction, resulting in low signal-to-noise ratio and spectrum leakage leads to time-domain reconstruction signal error.

Method used

The four-wave transverse shear interference phase recovery method based on the diamond Gaussian window function is adopted. The diamond Gaussian window function filtering process suppresses other stages of crosstalk and noise, reduces PV errors, and improves measurement accuracy.

Benefits of technology

It improves phase recovery accuracy, reduces the fine requirements of the optical path system, broadens the usage scenarios, is suitable for quantitative phase imaging of nano-scale transparent samples, reduces PV error, and improves the signal-to-noise ratio and smoothness of phase reconstruction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119085481A_ABST
    Figure CN119085481A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of optical measurement, in particular to a four-wave lateral shear interference phase restoration method based on a rhombus Gaussian window function. The method comprises the following steps: firstly, performing two-dimensional discrete Fourier transform on a four-wave lateral shear interferogram measurement graph, and filtering transformed frequency domain information by using a designed rhombus Gaussian window function to obtain positive and negative first-order frequency spectrums in x and y directions; then, positive or negative first-order frequency spectrums in the two directions are moved to a zero-order frequency spectrum center, and differential phases in the x direction and the y direction can be obtained through two-dimensional inverse discrete Fourier transform, background light removal, arc tangent operation and the like in sequence; and finally, performing least square fitting on the differential phases in the x and y directions by using a Fourier basis function, and substituting a Fourier coefficient obtained by fitting into the basis function again, so that a phase restoration result of the measured wavefront can be calculated. Compared with a reconstruction PV error obtained by adopting other common filtering window function phase extraction, the method provided by the invention can remove outliers which have great influence on a measurement result, so that the method has high phase restoration precision and contrast ratio.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of optical measurement technology, and in particular to a four-wave lateral shearing interference phase restoration method based on a rhombus Gaussian window function. Background Art

[0002] Real-time, high-precision phase restoration of nanoscale transparent samples can obtain the three-dimensional morphology of the samples, so it is widely used in the fields of quantitative phase imaging of living cells, surface detection of lenses and wafers. There are two common phase restoration methods, one is interferometry and the other is intensity analysis. As a type of interferometry, four-wave lateral shearing interferometry has the characteristics of achromatism, common optical path, high robustness, and high precision. It has been widely used in life medicine, integrated circuits, scientific research, and micro-nano manufacturing. The spatial resolution of four-wave lateral shearing interferometry is higher than that of the existing Shack-Hartmann sensor, and its accuracy is comparable to that of the existing interference-level holographic measurement technology. It is more compact than the traditional two-wave lateral shearing interferometry optical path and has the transient measurement advantage that the two-wave lateral shearing method cannot have.

[0003] The phase recovery method based on four-wave transverse shearing interference includes two parts: phase extraction and phase reconstruction. Commonly used phase extraction methods include time domain phase shift method, sinusoidal fitting method and Fourier transform method. The time domain phase shift method obtains the differential phase by introducing a phase difference between the test light and the reference light to obtain a set of interference pattern sequences. This method requires the collection of multiple interference patterns and is complicated to operate; the sinusoidal fitting method is an approximate method, which calculates the phase value at the midpoint position within each fringe cycle, and the accuracy is difficult to meet the measurement requirements of nano-scale samples; the Fourier transform method transforms the interference pattern signal from the time domain to the frequency domain, uses the filter window function to extract the positive or negative first-order spectrum containing the differential phase information, and then uses the inverse Fourier transform to obtain the differential phase information. However, in addition to extracting useful information, the commonly used rectangular filter window is usually mixed with noise and other level crosstalk, which reduces the signal-to-noise ratio of phase reconstruction. Summary of the invention

[0004] The present invention proposes a four-wave lateral shearing interferometry phase restoration method based on a diamond Gaussian window function. This method realizes phase extraction through a diamond Gaussian window function under the interference of noise and other miscellaneous frequencies. Compared with other commonly used filter window functions, this method can obtain higher phase restoration accuracy, reduce the fine requirements of the four-wave lateral shearing interferometry technology for the optical path system, and broaden the application scenarios. This method is mainly used for quantitative phase imaging of nanoscale transparent samples, and can accurately characterize samples to be tested with longitudinal dimensions on the order of wavelength.

[0005] The technical problems to be solved by the present invention are: 1. Most of the existing phase extraction methods for four-wave lateral shearing interferograms use rectangular filter windows to extract positive or negative first-order spectra, but in addition to the positive or negative first-order spectra containing phase information, the filter window is usually mixed with noise and other order crosstalk, and the signal-to-noise ratio of phase reconstruction is not high. 2. Since the four-wave lateral shearing interferogram actually obtained does not have strict periodicity, it is easy to truncate the signal when using a rectangular filter window to extract the phase, resulting in spectrum leakage. Accordingly, when using inverse Fourier transform to recover the signal, the truncation of the frequency domain signal will cause errors in the time domain reconstructed signal.

[0006] In view of the technical problems that the existing filter window is susceptible to noise, other-order spectrum crosstalk and spectrum leakage when extracting the phase in the four-wave lateral shearing interferometer pattern, the technical solution adopted by the present invention is: establishing a four-wave lateral shearing interferometer phase restoration method based on a diamond Gaussian window function, and extracting the differential phase through a diamond Gaussian window function according to the primary spectrum distribution characteristics of the transparent sample. The combined window function can suppress the influence of other-order crosstalk and noise, suppress spectrum leakage, reduce PV errors when measuring nanoscale transparent samples, and improve measurement accuracy.

[0007] The present invention provides a four-wave transverse shearing interference phase restoration method based on a rhombus Gaussian window function, comprising the following steps:

[0008] Step 1: Use a four-wave lateral shearing interferometry system to collect the carrier frequency interference pattern of the transparent sample. The light intensity distribution i(x,y) of the carrier frequency interference pattern is:

[0009]

[0010] Where (x, y) is the light intensity value of the camera pixel sampling point in the four-wave transverse shearing interferometry measurement system, I0 is the background light intensity of the interference pattern; fx' and fy' are the linear carrier frequencies introduced in the x and y directions respectively; and are the differential phase information in the x and y directions respectively; s is the shear amount; k is the wave number.

[0011] To facilitate Fourier transform, equation (1) is rewritten as:

[0012]

[0013] in:

[0014]

[0015]

[0016] Perform a two-dimensional discrete Fourier transform on equation (2) to obtain the spectrum distribution I(f x ,fy ):

[0017] I(f x ,f y )=A(f x ,f y )+C(f x -f x' ,f y )+C * (f x +f x' ,f y )+ D(f x ,f y -f y' )+D * (f x ,f y +f y' )(5)

[0018] In the formula, A(f x ,f y ) is the zero-order spectrum; C(f x -f x' ,f y ) and C * (f x +f x' ,f y ) is the positive and negative first-order spectrum in the x direction; D(f x ,f y -f y' ) and D * (f x ,f y +f y' ) are the positive and negative first-order spectra in the y direction.

[0019] Step 2: In order to eliminate the crosstalk of zero-order, second-order and other higher-order spectra on the phase reconstruction results, the obtained spectrum is filtered by combining the diamond window and Gaussian function to obtain a spectrum that only contains positive or negative first-order spectra in the x and y directions. The diamond Gaussian window function selects a diamond as the filter window shape in the horizontal direction and a Gaussian window as the filter function in the vertical direction. The Gaussian window is an exponential window with a wider main lobe, no negative side lobes, and a first side lobe attenuation of 55dB. The window range of the diamond Gaussian window function is:

[0020]

[0021] Where (u, v) is the frequency domain coordinate corresponding to (x, y), (u0, v0) is the center of the diamond filter window, r is the distance from the center point to the vertex of the diamond window, and the Gaussian window expression H(u, v):

[0022]

[0023] Where σ is the Gaussian distribution parameter, set σ = d / 3, d is the distance from the diamond vertex (u0, v0) through the point (u, v) to the diamond edge, then the diamond window contains 99.7% of the energy of the Gaussian window. Compared with other filter window functions, the combination of the diamond window shape and the Gaussian function avoids excessive invalid high-frequency information from participating in the subsequent phase reconstruction, thereby improving the overall phase contrast and reducing the PV error. Use the above diamond Gaussian window function to respectively convert the positive first-order spectrum C (f x -f x' ,f y ) and D(f x ,f y -f y' ) separated.

[0024] Step 3: Move the positive or negative first-order spectrum in both directions to the center of the spectrum graph to obtain the spectrum C(f x ,f y ) and D(f x ,f y ). Then, the two-dimensional inverse discrete Fourier transform is used to obtain c(x,y) and d(x,y) containing the differential phase information in the x and y directions:

[0025]

[0026] Step 4: Take out the sample, collect a four-wave lateral shearing interference pattern of the background light intensity, and repeat the above steps 1 to 3 under the same conditions of pixel size and number to obtain c′(x, y) and d′(x, y).

[0027] Step 5. Multiply the conjugate term c′(x,y) by (c′(x,y))*, and multiply d′(x,y) by (d′(x,y))* to obtain:

[0028]

[0029] in represents the differential phase after removing the background light intensity in the x direction, It represents the differential phase of removing the background light intensity in the y direction. This step can remove the dark current noise while removing the background light intensity of the optical path system.

[0030] Since C(f x ,f y ), D(f x ,f y ) are all known quantities. According to formula (9) and the inverse tangent operation, the differential phase of the background light intensity removed in the x and y directions can be obtained:

[0031]

[0032] in:

[0033]

[0034] Step 6: Based on the differential phases in the x and y directions obtained above, the original wavefront is reconstructed using the least squares Fourier transform method. It can be expanded using Fourier series:

[0035]

[0036] Where p and q are the frequency domain coordinates corresponding to x and y, M×M is the matrix dimension, and apq is the expansion coefficient of the Fourier series. and The wavefront to be measured is expressed as:

[0037]

[0038] In the formula, s is the shear amount. Substitute formula (13) into formula (12), introduce the deviation function, use the least squares method to fit, and when the deviation function is minimized, the expansion coefficient apq of the Fourier series is obtained:

[0039]

[0040] Where FT is the two-dimensional Fourier transform. From formula (12), it is easy to know that the phase to be measured is

[0041]

[0042] Beneficial effects of the present invention: The present invention provides a four-wave lateral shearing interference phase restoration method based on a diamond Gaussian window function, which combines the diamond window and the Gaussian function to filter the acquired spectrum, and can extract accurate differential phases, and provides a complete set of phase restoration methods from interference patterns to the wavefront to be measured for verification of the results. Compared with the common circular window shape and the combination of window functions such as rectangular windows, flat-top windows and Hanning windows, the phase PV error of the final reconstructed wavefront obtained by using the diamond Gaussian window function to extract the differential phase is smaller, the contrast is higher, and the surface is smoother. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of a four-wave transverse shearing interferometry phase recovery method based on a diamond Gaussian window function;

[0044] Figure 2 It is the spectrum obtained by Fourier transforming the four-wave transverse shearing interference pattern;

[0045] Figure 3 is the established diamond-shaped Gaussian window function;

[0046] Figure 4 It is the positive first-order spectrum in the x direction after being filtered by the diamond-shaped Gaussian window function;

[0047] Figure 5 It is the phase reconstructed by the four-wave transverse shearing interferometry phase restoration method based on the diamond Gaussian window function;

[0048] Figure 6 is the phase reconstructed using a rectangular window. DETAILED DESCRIPTION

[0049] The following will be combined Figure 1-6 An implementation case analysis is conducted on the technical solution of the present invention.

[0050] Step 1: First, use a warm white light source to emit a beam of visible light with a central band of 550nm through a quartz sample with nano-steps on the surface to obtain the wavefront to be measured, and then use an improved Shack-Hartmann mask (MHM) to divide the wavefront into four beams, and then place the camera sensor at the Talbot distance of the MHM to obtain a four-wave lateral shearing interference image. MHM can suppress ±3rd order light and even-order light, thereby obtaining four beams of ±1st order diffraction light with higher diffraction efficiency. The intensity distribution i(x,y) of the obtained interference pattern is:

[0051]

[0052] Where I0 is the background light intensity of the interference pattern; fx' and fy' are the linear carrier frequencies introduced by MHM in the x and y directions; and is the differential phase information in the x and y directions; s is the shear amount; k is the wave number.

[0053] The FFT algorithm of Matlab is used to perform two-dimensional discrete Fourier transform on equation (1):

[0054] I(f x ,f y )=A(f x ,f y )+C(f x -f x' ,f y )+C * (f x +f x' ,f y )+ D(f x ,f y -f y' )+D* (f x ,f y +f y' )(2)

[0055] In the formula, A(f x ,f y ) is the zero-order spectrum; C(f x -f x' ,f y ) and C * (f x +f x' ,f y ) is the positive and negative first-order spectrum in the x direction; D(f x ,f y -f y' ) and D * (f x ,f y +f y' ) are the positive and negative first-order spectra in the y direction. Figure 2 As shown, it is the spectrum diagram obtained after two-dimensional discrete Fourier transform. The zero-order spectrum is in the central area, and the four adjacent secondary spectra are the positive and negative first-order spectra in the x and y directions. It can be seen from the figure that there are many peaks near the first-order spectrum. These are invalid noise and high-frequency noise. When using the commonly used filter window function for phase extraction, it is inevitable to be affected by noise and noise. Take the center of the zero-order spectrum as the rectangular coordinate origin O, the horizontal right direction as the positive direction of the u axis, and the vertical upward direction as the positive direction of the v axis to establish the rectangular coordinate system uOv;

[0056] Step 2: Since the useful information of the primary spectrum is concentrated at the center of the primary spectrum and on the coordinate axis with the center of the primary spectrum as the origin, in order to eliminate the crosstalk effect of the zero-order, second-order and other higher-order spectra on phase extraction, a diamond is selected as the window shape of the filter window in the horizontal direction, which can ensure the extraction of effective information while suppressing the influence of invalid information; as shown in the attached figure Figure 3 As shown in the figure, a Gaussian window is selected as the window function of the filter window in the vertical direction. This function can suppress the influence of other "peaks" while ensuring full extraction of the primary spectrum. The Gaussian window is an exponential window with a wide main lobe, no negative side lobes, and a first side lobe attenuation of 55dB. Therefore, a diamond Gaussian combination window function is established to extract the differential phase. The window range of the combination window function is:

[0057]

[0058] Where (u, v) is the frequency domain coordinate corresponding to (x, y), (u0, v0) is the center of the diamond filter window, r is the distance from the center point to the vertex of the diamond filter window, and the window function expression H(u, v) of the combined window function is:

[0059]

[0060] Where σ is the Gaussian distribution parameter, set σ = d / 3, d is the distance from the center point (u0, v0) of the diamond to the edge of the diamond through the point (u, v), then the diamond window shape can contain 99.7% of the energy of the Gaussian window. Compared with the rectangular window shape, the diamond window shape avoids too much invalid high-frequency information from participating in the subsequent phase reconstruction, thereby improving the overall phase contrast and reducing the PV error. Use the above combined window function to respectively convert the positive first-order spectrum C (f x -f x' ,f y ), D(f x ,f y -f y' ) separated.

[0061] Step 3: Move the positive or negative first-order spectrum in both directions to the center of the spectrum graph to obtain the spectrum C(f x ,f y ), D(f x ,f y ), as attached Figure 4 As shown. Then use the two-dimensional discrete Fourier inverse transform to get c(x,y) and d(x,y).

[0062] Step 4: Take out the sample, collect a four-wave lateral shearing interference pattern of the background light intensity, and repeat the above steps 1 to 3 under the same conditions of pixel size and number to obtain c′(x, y) and d′(x, y).

[0063] Step 5. Multiplying the conjugate term c′(x,y) by (c′(x,y))* and d′(x,y) by (d′(x,y))* can remove the background light intensity and reduce the influence of dark current noise on phase reconstruction.

[0064] Since C(f x ,f y ), D(f x ,f y ) are all known quantities. The result of multiplying the conjugate terms is subjected to an inverse tangent operation to obtain the differential phase in the x and y directions after removing the background light intensity.

[0065]

[0066] Step 6: Based on the differential phases in the x and y directions obtained above, the original wavefront is reconstructed using the least squares-based Fourier transform method:

[0067]

[0068] Where FT and FT-1 are two-dimensional Fourier transform and inverse transform, is the phase of the wavefront to be measured. The optical path difference of the quartz sample is obtained by inverse Fourier transform of formula (6). Its phase is related to its own refractive index and the refractive index of air. The expression of the actual phase W(x, y) of the quartz step sample is:

[0069]

[0070] Where n1 is the refractive index of the quartz sample and nc is the refractive index of air.

[0071] According to the quartz sample interference pattern collected by the camera sensor, different combinations of window shapes and window functions are used to calculate the PV value after phase reconstruction. Two different window shapes, circular and diamond, are selected in the horizontal direction, and four different window functions, rectangular window, cylindrical inlaid cosine window, Hanning window and Gaussian window, are selected in the vertical direction, for a total of 8 combinations. The calculated PV results are shown in Table 1:

[0072] Table 1 PV of reconstructed phase under different combinations (λ=550nm)

[0073] The PV of the quartz sample measured by white light interferometer is less than 0.2000λ, so the smaller the PV value of the reconstructed phase, the smaller the error. According to Table 1, the reconstructed PV errors obtained by extracting the differential phase using a rhombus as the filter window shape are all smaller than those of a circle. The reconstructed PV errors obtained by extracting the differential phase using different window functions are from large to small: rectangular window>cylindrical inlaid cosine window>Hanning window>Gaussian window. Therefore, the reconstructed PV error obtained by extracting the differential phase using a rhombus Gaussian window is the smallest, indicating that the effect of filtering out clutter and noise is better, and it is more suitable for wavefront phase reconstruction of four-wave lateral shearing interferograms.

[0074] Figure 5 It is the local wavefront quality of the reconstructed phase after extracting the differential phase using a circular window shape and a rectangular window function as a combined window function.

[0075] Figure 6 It is the local wavefront quality of the reconstructed phase after extracting the differential phase using the diamond Gaussian window function as the combined window function.

[0076] It is easy to know that the use of a diamond Gaussian window function can remove outliers with greater influence, reduce the impact of noise, and reconstruct a smoother and flatter surface of the phase. The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments, but the scope of implementation of the present invention is not limited thereto.

Claims

1. A four-wave transverse shearing interferometry phase recovery method based on a diamond Gaussian window function. Features The following steps are involved: Step 1: Use a four-wave lateral shearing interferometry measurement system to collect the carrier frequency interference pattern of the sample, and then use a two-dimensional discrete Fourier transform algorithm to convert it into a spectrum diagram, where the zero-order spectrum is in the central area, and the four adjacent secondary spectra are the positive and negative first-order spectra in the x and y directions respectively; Step 2: In order to eliminate the crosstalk effect of zero-order, second-order and other higher-order spectra on the phase reconstruction result, the obtained spectrum is filtered by combining the diamond window and Gaussian function to obtain a spectrum containing only the positive or negative first-order spectra in the x and y directions; Step 3: After moving the positive or negative first-order spectrum in both directions to the center of the spectrum diagram, the differential phase in the x and y directions is obtained by using a two-dimensional discrete Fourier inverse transform; Step 4: Take out the sample, collect the four-wave lateral shearing interference pattern of the background light intensity, and use the above method to obtain the differential phase of the background light intensity in the x and y directions under the same conditions of pixel size and number; Step 5: Subtract the differential phase corresponding to the background light intensity measurement from the two-directional differential phase corresponding to the sample measurement to obtain the differential phase in the x and y directions after removing the background light; Step 6: Use the Fourier basis function to perform least squares fitting on the differential phase in the x and y directions after removing the background light, and substitute the Fourier coefficients obtained by fitting into the basis function again to calculate the phase restoration result of the measured wavefront.