Intelligent adaptive wavefront measurement system and method for optical characterization
Through the intelligent adaptive wavefront measurement system, a shear generator is formed using a spatial light modulator and a phase difference microlens array, combined with machine learning to optimize the modulation mode, the problem of low multi-parameter measurement accuracy in traditional optical characterization technology is solved, and the problem of low multi-parameter measurement accuracy is realized High precision and sensitivity optical parameter detection.
Patent Information
- Application Number
- PCT/CN2023/094445
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-16
- Publication Date
- 2024-11-21
AI Technical Summary
Existing optical characterization technology is difficult to measure multiple optical parameters with high accuracy in a single system. The traditional wavefront measurement method is limited by reference mirror processing accuracy and microlens manufacturing technology, resulting in low detection accuracy and energy utilization, which cannot meet the needs of complex optical systems. High precision requirements.
Adopting an intelligent adaptive wavefront measurement system, a shear generator is formed through a spatial light modulator and a phase difference microlens array, and using machine learning to optimize the modulation mode to realize automatic modulation and wavefront reconstruction, eliminate interference graph noise, and improve measurement accuracy and sensitivity.
It extends the application range of wavefront measurement, improves the accuracy and sensitivity of optical parameter measurement, and realizes fast and convenient optical characterization, is suitable for online measurement, has strong anti-environmental interference capability, and is suitable for the detection of a variety of optical parameters.
Smart Images

Figure CN2023094445_21112024_PF_FP_ABST
Abstract
Description
Intelligent adaptive wavefront measurement system and method for optical characterization Technical Field
[0001] The present invention relates to an intelligent adaptive measurement system and method for optical characterization. Background Art
[0002] The continuous progress of new optical design methods, new materials and new systems has greatly promoted the innovative research and development of new products, but at the same time has also put forward higher requirements for their optical characterization. Optical characterization refers to the use of optical parameters to determine the acceptability of optical systems or optical components under specific performance requirements. It is widely needed in different fields, such as the power distribution of eyeglass lenses in ophthalmic optics, the quality of laser beams in two-photon 3D printing, the refractive index distribution in materials science, the assembly error of high numerical aperture objectives, the focal length of metalenses, the optical transfer function (OTF), the modulation transfer function (MTF), etc. Optical characterization is a multi-parameter, complex measurement process. Although there are some equipment and instruments that can measure a single optical parameter in a specific field, there is still a lack of methods and equipment that can accurately measure multiple optical parameters in a single system as a whole.
[0003] Wavefront measurement is an effective method for optical characterization. It primarily defines the deviation between the actual and ideal wavefronts. It detects the phase change of a measuring beam after it passes through different media or components, leveraging the principle of interference to perform non-destructive probing of the object under test. Figure 1 shows a schematic diagram of an optical system 100 utilizing wavefront measurement methods in the prior art. Traditional wavefront measurement methods include the point diffraction method, the Shack-Hartmann method, and the four-wave transverse shear method.
[0004] Traditional interferometry detection relies on a standard mirror in the optical path to generate a reference wavefront, which not only limits the detection aperture of the interferometry system, but also makes it difficult to generate a high-quality reference spherical wavefront due to the limited processing accuracy of the standard reference mirror, making it impossible to perform high-precision optical detection. The point diffraction method uses a small hole with a size close to the detection wavelength to generate a nearly ideal spherical wave as the reference wavefront in the interferometry detection system, thereby achieving high-precision interferometry detection. Specifically, the point diffraction method diffracts the convergent wavefront to be measured through the small hole in the point diffraction plate to produce a nearly ideal reference spherical wave, and the wave directly transmitted from the point diffraction plate is the test light wave carrying the information of the wavefront to be measured. The two interfere with each other, and the phase distribution of the wavefront to be measured can be demodulated from the interference pattern. The point diffraction method does not require a reference mirror, so it has strong resistance to environmental interference. However, the processing accuracy of the small hole in the diffraction plate is very high, and the light beam alignment is difficult. At the same time, the blocking effect of the small hole on the measurement beam makes the interference pattern low in contrast.
[0005] The Hartmann method uses a regularly arranged array of holes as shown in Figure 2A as an aperture, controlling the transmission of light at different positions through the aperture array, ultimately causing it to be incident on the receiving screen at different slopes. However, such an aperture blocks most of the energy, making the final light spot difficult to measure. The Shack-Hartmann method improves the Hartmann aperture into a microlens array, as shown in Figure 2B. Although the Shack-Hartmann method greatly improves the utilization rate of light energy, it is constrained by microlens manufacturing technology (the size and number of microlenses are limited), resulting in a limited number of sampling points for the wavefront to be measured, resulting in a low spatial resolution of this method.
[0006] The four-wave transverse shearing method uses a checkerboard grating (as shown in Figure 2C) as a spectroscopic element. After the light wave passes through the grating, a replica light wave is obtained that has a certain lateral displacement from itself. This replica light wave interferes with the light wave to be measured, forming transverse shear interference, and interference fringes appear where the two overlap. The interference fringes contain differential information of the original wavefront, and the original wavefront can be reproduced through specific analysis and quantitative calculation (inverse Fourier transform). The four-wave transverse shearing method greatly simplifies the system structure, but there are still problems such as the influence of high-order diffraction and the fact that high-precision measurement can only be achieved at positions that are multiples of 1 / 6 of the Talbot distance.
[0007] Summary of the Invention
[0008] In order to overcome the above problems, the present invention proposes an adaptive intelligent wavefront measurement method and system. After the spherical wave or plane wave emitted from the light source passes through the system or component to be measured, it is deformed. The distorted wavefront is copied into multiple wavefronts to be measured by a specially designed shear generator, and interferes with each other. The charge coupled device (CCD) records the interference pattern formed by them. The interference pattern is reconstructed to obtain the wavefront information to be measured, which is converted into various optical parameters to be measured through a mapping method. The shear generator consists of a spatial light modulator (SLM) and a microlens array with phase difference (OPD-MLAs). The spatial light modulator is automatically modulated based on the adaptive technology of machine learning to meet the needs of different scenes to be measured. The present invention expands the application scope of wavefront measurement, improves the accuracy of wavefront measurement, and makes fast and convenient optical characterization possible.
[0009] The present invention proposes a novel intelligent adaptive measurement system for optical characterization. It can be widely used in optical parameter measurement, such as beam quality evaluation, optical component detection, metasurface, material refractive index and laser atmospheric communication and human eye aberration measurement. Compared with traditional interferometers, it does not require reference light. It uses a shear generator to self-replicate the wavefront to be measured and then generate interference. The interference pattern recorded by the CCD contains the wavefront information of each shear direction. The phase information of the sample to be measured can be obtained by using wavefront reconstruction technology. It has a simple structure, strong resistance to environmental interference, and is suitable for online measurement. Adaptive technology is applied to the wavefront modulation of the shear generator, and has the ability to automatically adjust key shear parameters (such as duty cycle, shear amount and illumination intensity, etc.) to meet the needs of different application scenarios; the interference pattern obtained has low noise, strong contrast and is stable. The present invention expands the application range of wavefront measurement and improves the accuracy of wavefront measurement.
[0010] Specifically, the present invention discloses an intelligent adaptive wavefront measurement system for optical characterization, which can obtain the optical parameters of the sample to be measured. The system includes a light source, a shear generator and a photoelectric coupling device arranged in sequence along the axial direction, wherein the shear generator includes a spatial light modulator and a phase difference microlens array arranged in sequence along the axial direction.
[0011] In some embodiments, the phase difference microlens array includes convex microlenses and rear convex microlenses arranged alternately on adjacent grids. The phase difference microlens array performs π phase modulation on the incident light wave through the axial height difference between the top of the convex microlenses and the top of the rear convex microlenses.
[0012] In some embodiments, in the subsequent wavefront reconstruction operation, the light transmittance of the top of the front convex microlens is calculated as 1, the light transmittance of the top of the base of the phase difference microlens array is calculated as 0, and the light transmittance of the top of the rear convex microlens is calculated as -1, where the height difference is Wherein, n is the refractive index of the material of the phase difference microlens array, and λ is the incident wavelength.
[0013] In some embodiments, the duty cycle of each microlens in the phase difference microlens array in its grid is Where r is the aperture radius of each microlens, and d is the grid spacing.
[0014] In some embodiments, the duty cycle is 2:3.
[0015] In some embodiments, the shear ratio of the shear generator is Among them, S is the shear amount, D is the aperture of the incident light, among them, L is the distance between the photoelectric coupling device and the phase difference microlens array, H is the thickness of the substrate of the phase difference microlens array, λ is the incident wavelength, and d is the grid spacing.
[0016] In some embodiments, the spatial light modulator modulates a shearing parameter, where the shearing parameter includes at least one of the following: a duty cycle of each microlens of the phase-difference microlens array in its grid, a shearing amount, and an illumination intensity.
[0017] In some embodiments, the spatial light modulator optimizes the modulation mode of the spatial light modulator based on a machine learning model, so that the machine learning model can output a modulation mode that is adaptive to the sample to be tested based on the spectrum signal under the initial modulation mode.
[0018] In some embodiments, the machine learning model is established using a deep convolutional neural network.
[0019] In some embodiments, the machine learning model can optimize the modulation mode of the spatial light modulator so that the spectrum signal corresponding to the wavefront light intensity distribution received on the optoelectronic coupling device only has the 0th order and ±1st order.
[0020] In some embodiments, the training process of the machine learning model can be carried out in a simulation environment. By designing a sample to be tested with a complex structure, the optical parameters of the designed structure are obtained in the simulation. Since the optical parameters are known, the optical parameters are used as the true value of the deep convolutional neural network training, and through continuous iterative learning and training, a machine learning model that has completed the training is finally obtained.
[0021] In addition, the present invention also discloses an intelligent adaptive wavefront measurement method for optical characterization, which includes the following steps:
[0022] a) An incident light wave from a light source passes through a sample to be measured, a shear generator, and a photoelectric coupling device sequentially arranged along an axial direction to obtain a wavefront light intensity distribution, wherein the shear generator includes a spatial light modulator and a phase difference microlens array sequentially arranged along an axial direction;
[0023] b) After the wavefront light intensity distribution is subjected to Fourier transform, a spectrum signal corresponding to the wavefront light intensity distribution is obtained;
[0024] c) performing inverse Fourier transform and shearing interferometry signal processing on the spectrum signal to obtain the optical parameters of the sample to be measured.
[0025] In some embodiments, the phase difference microlens array includes convex microlenses and rear convex microlenses arranged alternately on adjacent grids. The phase difference microlens array performs π phase modulation on the incident light wave through the axial height difference between the top of the convex microlenses and the top of the rear convex microlenses.
[0026] In some embodiments, the light transmittance of the front convex microlens is 1, the light transmittance of the bottom portion is 0, and the light transmittance of the rear convex microlens is -1, wherein the height difference is Wherein, n is the refractive index of the material of the phase difference microlens array, and λ is the incident wavelength.
[0027] In some embodiments, the duty cycle of each microlens in the phase difference microlens array in its grid is Where r is the aperture radius of each microlens, and d is the grid spacing.
[0028] In some embodiments, the duty cycle is 2:3.
[0029] In some embodiments, the shear ratio of the shear generator is Among them, S is the shear amount, D is the aperture of the incident light, among them, L is the distance between the photoelectric coupling device and the phase difference microlens array, H is the thickness of the substrate of the phase difference microlens array, λ is the incident wavelength, and d is the grid spacing.
[0030] In some embodiments, the spatial light modulator modulates a shearing parameter, wherein the shearing parameter includes at least one of the following: a duty cycle of each microlens of the phase difference microlens array in its grid, a shearing amount, and an illumination intensity.
[0031] In some embodiments, the spatial light modulator optimizes the modulation mode of the spatial light modulator based on a machine learning model, so that the machine learning model can output a modulation mode that is adaptive to the sample to be tested based on the spectrum signal under the initial modulation mode.
[0032] In some embodiments, the machine learning model is established using a deep convolutional neural network.
[0033] In some embodiments, the machine learning model can optimize the modulation mode of the spatial light modulator so that the spectrum signal corresponding to the wavefront light intensity distribution received on the optoelectronic coupling device only has the 0th order and ±1st order.
[0034] In some embodiments, the training process of the machine learning model can be carried out in a simulation environment. By designing a sample to be tested with a complex structure, the optical parameters of the designed structure are obtained in the simulation. Since the optical parameters are known, the optical parameters are used as the true value of the deep convolutional neural network training, and through continuous iterative learning and training, a machine learning model that has completed the training is finally obtained.
[0035] In addition, the present invention also discloses a training method for an intelligent adaptive wavefront measurement method for optical characterization in a simulation environment, the method comprising the following steps:
[0036] a) In a simulation environment, an incident light wave from a light source passes through a sample with known optical parameters, a spatial light modulator, a phase-difference microlens array, and an optoelectronic coupling device arranged in sequence along the axial direction to obtain an initial wavefront intensity distribution, wherein the spatial light modulator operates in an initial modulation mode;
[0037] b) After the initial wavefront light intensity distribution is subjected to Fourier transform, an initial spectrum signal corresponding to the initial wavefront light intensity distribution is obtained;
[0038] c) Inputting the initial spectrum signal and the current modulation mode into a machine learning model built using a deep convolutional neural network, extracting and reconstructing features of the input signal through deep convolution, and outputting the adjusted modulation mode;
[0039] d) Calculate the remaining level of the spectrum signal at this time and include it in the training loss for model training;
[0040] e) obtaining an adjusted wavefront light intensity distribution based on the adjusted modulation mode, and obtaining an adjusted spectrum signal corresponding to the adjusted wavefront light intensity distribution after Fourier transforming the adjusted wavefront light intensity distribution;
[0041] f) performing inverse Fourier transform and shearing interferometry signal processing on the adjusted spectrum signal to obtain the optical parameters of the sample, and calculating the error between the obtained optical parameters and the true values of the known optical parameters, and accounting for the error in the training loss;
[0042] g) Iteratively train the model based on the obtained training loss, and finally obtain a fully trained machine learning model.
[0043] In addition, the present invention also discloses an intelligent adaptive wavefront measurement method for optical characterization, which includes the following steps:
[0044] a) An incident light wave from a light source passes through a sample to be measured, a spatial light modulator, a phase-difference microlens array, and a photoelectric coupling device arranged in sequence along an axial direction to obtain an initial wavefront intensity distribution, wherein the spatial light modulator operates in an initial modulation mode;
[0045] b) After the initial wavefront light intensity distribution is subjected to Fourier transform, an initial spectrum signal corresponding to the initial wavefront light intensity distribution is obtained;
[0046] c) Inputting the initial spectrum signal and the current modulation mode into the trained machine learning model, extracting and reconstructing features of the input signal through deep convolution, and adjusting the modulation mode so that the spectrum signal corresponding to the wavefront intensity distribution received by the optoelectronic coupling device only has levels 0 and ±1;
[0047] d) adjusting the modulation mode of the spatial light modulator based on the adjusted modulation mode to obtain an optimized wavefront intensity distribution, and performing Fourier transform on the optimized wavefront intensity distribution to obtain an optimized spectrum signal corresponding to the optimized wavefront intensity distribution;
[0048] e) performing inverse Fourier transform and shearing interferometry signal processing on the optimized spectrum signal to obtain the optical parameters of the sample to be measured.
[0049] Other features and advantages will become apparent from the following detailed description, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] The foregoing and further features of the present invention will become apparent from the following description of preferred embodiments provided by way of example only and taken in conjunction with the accompanying drawings, in which:
[0051] FIG1 is a schematic structural diagram of an optical system using a wavefront measurement method in the prior art;
[0052] FIG2A shows a schematic diagram of a Hartmann diaphragm in the prior art;
[0053] FIG2B shows a schematic diagram of a microlens array in the Shack-Hartmann method in the prior art;
[0054] FIG2C shows a schematic diagram of a checkerboard grating in a four-wave transverse shearing method in the prior art;
[0055] FIG3 shows a schematic diagram of an intelligent adaptive wavefront measurement system for optical characterization according to an embodiment of the present invention;
[0056] FIG4 is a schematic diagram showing a shearing interference wavefront received by a CCD camera according to an embodiment of the present invention;
[0057] FIG5 is a schematic diagram showing a wavefront reconstruction algorithm according to an embodiment of the present invention;
[0058] FIG6 is a schematic diagram showing a wavefront phase distribution to be measured according to an embodiment of the present invention;
[0059] 7A-7C are schematic diagrams showing a phase difference microlens array according to an embodiment of the present invention;
[0060] FIG7D is a schematic diagram showing the incident light aperture and shearing amount according to an embodiment of the present invention;
[0061] FIG7E is a schematic diagram showing the distance between a CCD camera and a phase-difference microlens array according to an embodiment of the present invention;
[0062] FIG8 shows a schematic diagram of wavefront modulation based on machine learning according to an embodiment of the present invention;
[0063] FIG9 shows a method for self-optimization of shearing interferometry modulation patterns based on machine learning according to an embodiment of the present invention;
[0064] FIG10 is a schematic flow chart showing a training method of an intelligent adaptive wavefront measurement method for optical characterization in a simulation environment according to some embodiments of the present invention;
[0065] FIG11 is a schematic flow chart showing an intelligent adaptive wavefront measurement method for optical characterization according to some embodiments of the present invention;
[0066] FIG12A shows an unmodulated spectrum diagram according to an embodiment of the present invention;
[0067] FIG12B shows a spectrum diagram after modulation according to an embodiment of the present invention. DETAILED DESCRIPTION
[0068] In the following claims and the foregoing description of the invention, unless the context requires otherwise due to expressive language or necessary implication, the word "comprise" or variations such as "include" or "contains" are used in an inclusive manner, that is, to specify the presence of the features described, but not to exclude the presence or addition of additional features in various embodiments of the invention. In addition, unless otherwise specified, the dimensions and proportions in the drawings do not represent the actual dimensions and proportions of the device.
[0069] It will be appreciated that, if any prior art publication is cited herein, such reference does not constitute an admission that the publication forms part of the common general knowledge in the art in any country.
[0070] FIG3 shows a schematic diagram of an intelligent adaptive wavefront measurement system 300 for optical characterization according to an embodiment of the present invention. The system 300 includes a light source 310, a sample to be measured 320, a shear generator 330, and a CCD camera 360. The aforementioned devices can be arranged sequentially along the axis z. The light source 310 is used to generate, for example, a standard plane wave or a spherical wave. The light source can be a laser diode light source, and the wavelength range can be, for example, 380 nm to 1200 nm, preferably in the visible light band. The shear generator 330 is composed of a spatial light modulator 340 and a phase-difference microlens array 350, which will be described in detail later. The shear generator 300 acts as a spectroscopic element to simultaneously replicate the wavefront to be measured into multiple wavefronts. Since the replicated wavefronts have different spatial exit angles, they are offset from each other on the observation plane of the CCD camera 360, forming shear interference wavefronts in multiple directions. FIG4 shows a schematic diagram of the shear interference wavefronts received on the CCD camera according to an embodiment of the present invention. The CCD camera 360 is merely exemplary and can also be other devices that convert or couple optical signals into electrical signals. System 300 utilizes a wavefront reconstruction algorithm, as shown in Figure 5, to capture lateral shearing interference information in multiple directions through a single interferogram. This allows measurement of the wavefront phase distribution to be measured, as shown in Figure 6, which illustrates a schematic diagram of the wavefront phase distribution to be measured.
[0071] The most critical optical component in the wavefront measurement system is the grating as a spectrometer. After the light wave to be measured passes through the grating, multiple diffraction orders will be generated due to the diffraction effect of the grating. By designing a special grating structure, in addition to the four diffraction orders of ±1 in two orthogonal directions that participate in interference, the remaining diffraction orders should be eliminated as much as possible, thereby avoiding the influence of stray diffraction light on the detection accuracy of the system. The ideal wavefront transverse shear grating Fraunhofer diffraction field distribution can be expressed as follows:
[0072] Among them, u = λ / x and v = λ / y are the frequency domain coordinates corresponding to the spatial coordinates x and y respectively, λ is the wavelength of the light source, and d is the spacing between adjacent grids of the grating, that is, the grating period (see Figure 7C). The δ function is the Dirac function, that is, a generalized function or distribution defined on the real line, which is equal to zero at all points except zero, and the integral over the entire domain is equal to 1. According to the Fraunhofer diffraction theory, the field distribution on the observation surface is the Fourier transform of the field distribution of the diffraction grating transmitted light. When an ideal plane wave is incident vertically on the grating, the grating transmittance distribution function can be obtained by Fourier inverse transform of formula (1), that is
[0073] From formula (2), we can see that the transmittance of the transverse shear grating ranges from [-1, 1]. The amplitude grating cannot achieve a negative refractive index, so we consider rewriting the above formula as the product of the absolute value of the transmittance distribution function and a positive and negative sign function, t ideal (x,y)=sgn[t ideal (x,y)]·abs[t ideal (x,y)] (3)
[0074] The positive and negative sign functions can be easily realized by a phase grating with a periodic alternating distribution of phase delays of 0 and π. The phase delay can be expressed as the convolution of the rectangular function in the x and y directions with the comb function, that is,
[0075] From the above formula, we can see that the grating transmittance at the position where the phase delay is π is -1, and the grating transmittance at the position where the phase delay is 0 is 1, so the ideal shear grating transmittance distribution is
[0076] Figures 7A-7C illustrate schematic diagrams of a phase-difference microlens array 700 according to an embodiment of the present invention. In some embodiments of the present invention, as an alternative to an ideal grating with continuously varying transmittance, the phase-difference microlens array 700 phase-modulates the incident light wave by utilizing the height difference between adjacent microlenses in the spatial position. Specifically, the phase-difference microlenses modulate the incident light wave by a factor of π by utilizing the height difference between adjacent microlenses in the axial direction z. Figure 7A shows convex microlenses 710 and convex microlenses 720 arranged alternately on adjacent grids. The phase-difference microlens array 700 of the present invention can be an m*m grid array (m is a positive integer greater than or equal to 2). Figure 7B schematically illustrates a 7*7 grid array. Figure 7C illustrates each grid 750 in the phase-difference microlens array 700 (e.g., as shown by the dashed box in Figure 7B). As can be seen from Figures 7A-7C , the grid spacing is d. The microlenses are arranged in the center of the grid, and their projection onto a plane perpendicular to the axial direction z is approximately a circle with a radius of r. The phase difference microlens array 700 includes a substrate 730 with a thickness of H, a convex microlens 710 and a convex microlens 720 with a height difference of h formed in the axial direction z. The convex microlens 710 and the convex microlens 720 in the present invention can be a microlens array formed of the same material as the substrate 730. In a preferred embodiment of the present invention, the convex microlens 710 and the convex microlens 720 both protrude in the opposite direction of the z axis. The convex microlens 710 and the convex microlens 720 are staggered on the grid array shown in Figure 7B. That is, each convex microlens 710 is surrounded by the convex microlens 720 in the up, down, left and right directions (unless it reaches the edge of the grid array), and each convex microlens 720 is surrounded by the convex microlens 710 in the up, down, left and right directions (unless it reaches the edge of the grid array). Phase-difference microlens array 700 utilizes material refraction to achieve continuous variation in the transmittance of the incident wave. It also utilizes the height difference h between the top 712 of the front convex microlens 710 and the top 722 of the rear convex microlens 720 in the axial direction z to achieve phase modulation of π. The present invention does not preclude the use of other devices, such as apertures, gratings, or microlens arrays, to achieve phase modulation of π. For example, phase modulation can be achieved by using different materials, incident angles, or sizes of adjacent gratings.
[0077] 7A , in the subsequent wavefront reconstruction operation, the light transmittance of the top 712 of the front convex microlens 710 is calculated to be 1. The light transmittance of the top 732 of the substrate 730 is calculated to be 0. The light transmittance of the top 722 of the rear convex microlens 720 is calculated to be -1. The height difference h between the top 712 of the front convex microlens 710 and the top 722 of the rear convex microlens 720 in the axial direction z is
[0078] Where n is the refractive index of the material of the phase difference microlens array, and λ is the incident wavelength. In order to better eliminate the influence of even-order light and ±3rd-order diffracted light on the measurement, the duty ratio of each microlens in the phase difference microlens array 700 in its grid is 2:3. Combined with Figure 7C, the duty ratio can be expressed as
[0079] Where r is the aperture radius of each microlens, and d is the grid spacing. Referring to Figures 7D-7E, Figure 7D shows the incident light aperture D and the shear amount S, and Figure 7E shows the distance L between the CCD camera and the phase difference microlens array. Note that the following parameters are important parameters that affect the system shear rate, such as the thickness H of the substrate 730 of the phase difference microlens array, the aperture D of the incident light, and the distance L between the CCD camera 360 and the phase difference microlens array 700. From the geometric relationship between Figures 7D and 7E, it can be seen that the shear amount S is
[0080] Then the shear ratio β is
[0081] The smaller the shear ratio, the smaller the distance between wavefronts, the more wavefronts participate in interference, the larger the measurable aperture, and the higher the accuracy of wavefront reconstruction. Conversely, the larger the shear ratio, the larger the distance between wavefronts, the smaller the wavefronts participate in interference, and the higher the measurement sensitivity, but the measurable aperture decreases, and the corresponding wavefront reconstruction accuracy decreases. Therefore, in practical applications, the shear ratio of multi-beam lateral shearing interferometry technology requires a high degree of adjustable freedom. The function of achieving adjustable shear ratio by adjusting spatial light modulator 340 will be described in detail below.
[0082] Spatial light modulators are suitable for adjusting key shear parameters. They can be traditional optical devices or, for example, liquid crystal arrays, where the transmittance of the liquid crystals is varied by adjusting voltage. This paper introduces machine learning-based adaptive technology into a spatial light modulator for automatically modulating wavefront measurement, achieving automatic modulation of wavefront measurement.
[0083] Figure 8 illustrates a schematic diagram of wavefront modulation based on machine learning according to some embodiments of the present invention. In some embodiments of the present invention, the machine learning-based adaptive spatial light modulator 340 can automatically adjust key shearing parameters, such as duty cycle, shearing amount, and illumination intensity, to reduce the diffraction spectrum to only the 0th and ±1st orders.
[0084] Figure 9 shows a method for self-optimization of shearing interferometry modulation patterns based on machine learning according to some embodiments of the present invention. The machine learning model can output a modulation pattern that is adaptive to the sample to be tested based on the spectrum signal under the initial modulation pattern.
[0085] First, under the initial modulation mode, an initial wavefront intensity distribution is obtained. After Fourier transform (FT), the corresponding spectrum signal of this initial wavefront is obtained. This initial spectrum signal is input into a machine learning-based model for inference, obtaining a modulation mode suitable for the sample under test. The modulation mode of the spatial light modulator is then optimized to obtain an optimized spectrum signal. Finally, through inverse Fourier transform (IFT) and shearing interferometry signal processing, the optical parameters of the sample under test are determined.
[0086] In this intelligent adaptive process, the machine learning model will be established using a deep convolutional neural network, which can process the input two-dimensional spectrum signal, extract and reconstruct the input signal features through the deep convolution layer, and finally output the optimized modulation mode.
[0087] Preferably, the training process of the learning model can be carried out in a simulation environment. By designing a sample to be tested with a complex structure, the optical parameters of the designed structure are obtained in the simulation. Since these parameters are known, they can be used as the true values for training the deep convolutional network model. Through continuous iterative learning and training, a fully trained machine learning model is ultimately obtained. The deep convolutional network model can output a modulation pattern that is adaptive to the sample to be tested based on the spectrum signal under the initial modulation pattern.
[0088] In order to simulate the diffraction effect of the shear generator 300, a plane wave is incident to generate Fraunhofer diffraction, and the diffraction field distribution is:
[0089] where t(x,y) is the grating transmittance distribution, F[·] represents the Fourier transform, and the wave number It represents the phase change produced per unit length, λ is the wavelength of the light source, and L is the distance between the CCD camera and the phase difference microlens array.
[0090] FIG10 illustrates a flow chart of a training method for an intelligent adaptive wavefront measurement method for optical characterization in a simulation environment according to some embodiments of the present invention. First, in step 1010, referring to the system shown in FIG3 , a light source 310 passes through a sample 320 with known optical parameters and then through a shear generator 330. In step 1020, an initial wavefront intensity distribution is obtained at a CCD camera 360. The shear generator 330 includes a spatial light modulator 340 in an initial modulation mode and may also include a phase-difference microlens array 350 or any other optical device capable of generating a phase difference of π. In step 1030, a Fourier transform is performed on the initial wavefront intensity distribution using any computing device known in the art. In step 1040, an initial spectral signal corresponding to the initial wavefront intensity distribution is obtained. In step 1050, the spectral signal and modulation mode are used as inputs, and a machine learning model is used to extract and reconstruct features of the input signal through deep convolution to obtain an adjusted modulation mode. In step 1060, the remaining levels of the spectral signal are calculated and factored into the training loss. In step 1070, the spectrum signal undergoes inverse Fourier transform and shearing interferometry signal processing. In step 1080, the error between the obtained sample optical parameters and the known true values of the sample parameters is calculated and factored into the training loss. In step 1090, it is determined whether the training loss is close to zero or the maximum number of training iterations has been reached. If so, the process proceeds to step 1100; otherwise, the process returns to step 1050. In step 1100, the trained machine learning model is output.
[0091] FIG11 illustrates a flow chart of an intelligent adaptive wavefront measurement method for optical characterization according to some embodiments of the present invention. First, in step 1110, referring to the system shown in FIG3 , a light source 310 passes through a sample to be measured 320 and then through a shear generator 330. In step 1120, an initial wavefront intensity distribution is obtained at a CCD camera 360. The shear generator 330 includes a spatial light modulator 340 operating in an initial mode and may also include a phase-difference microlens array 350 or any other optical device capable of generating a phase difference of π. In step 1130, a Fourier transform is performed on the initial wavefront intensity distribution using any computing device known in the art. In step 1140, an initial spectral signal corresponding to the initial wavefront intensity distribution is obtained. In step 1150, the initial spectral signal and the current modulation mode are input into a trained machine learning model. Deep convolution is used to extract and reconstruct features from the input signal. The modulation mode is adjusted so that the spectral signal corresponding to the wavefront intensity distribution received by the photoelectric coupling device only contains levels 0 and ±1. At step 1160, the modulation mode of the spatial light modulator is adjusted based on the adjusted modulation mode to obtain an optimized wavefront intensity distribution. The optimized wavefront intensity distribution is Fourier transformed to obtain an optimized spectrum signal corresponding to the optimized wavefront intensity distribution. At step 1170, the optimized spectrum signal is subjected to an inverse Fourier transform and shearing interferometer signal processing to obtain the optical parameters of the sample to be measured.
[0092] Please note that the processes and step sequences in Figures 10 and 11 are illustrative and non-restrictive, and those skilled in the art can adjust or modify the steps and sequence based on feasibility.
[0093] FIG12A shows an unmodulated spectrum diagram according to some embodiments of the present invention, wherein the abscissa represents the spatial frequency, with a unit of 1 / d, where d is the grid spacing, and the ordinate represents the normalized intensity of the diffracted field.
[0094] Figure 12B shows a modulated spectrum according to some embodiments of the present invention. The horizontal axis represents spatial frequency, with units of 1 / d, where d is the grating spacing, and the vertical axis represents the normalized intensity of the diffracted field. As can be seen, the modulated spectrum only contains light of order 0 (not shown) and ±1, while the majority of higher-order light is eliminated. Compared to commonly used modified Hartmann masks (MHMs) and phase gratings, the diffraction effect is closer to that of an ideal grating.
[0095] The present invention can be widely used in optical parameter measurement, such as beam quality evaluation, optical component detection, metasurface, material refractive index and laser atmospheric communication and human eye aberration measurement.
[0096] Compared with the existing technology, the present invention addresses the lack of integrated characterization methods for multiple optical parameters and proposes a novel intelligent adaptive measurement system, which can be widely used in various wavefront measurement scenarios, such as laser beam quality, lens refractive power distribution, metamaterials, precision lens assembly, material refractive index distribution, etc.
[0097] To address the problem of measurement scene differentiation, the present invention proposes that an adaptive technology based on machine learning will be used for wavefront modulation of the spatial light modulator. This method can accurately and automatically modulate the incident wavefront into an arbitrary wavefront, which provides a basis for the integration of the measurement system.
[0098] In response to the influence of high-order diffracted light in shearing interference, the spatial light modulator and phase-difference microlens array proposed in the present invention are combined into a shear generator, which can achieve close to single-order diffraction and greatly improve the sensitivity and resolution of the system.
[0099] The present invention is shown and described in detail above with reference to the accompanying drawings, but the above embodiments should be considered as illustrative rather than restrictive. The present invention also includes various combinations, modifications and variations of the exemplary embodiments without departing from the spirit and scope of the present invention.
Claims
1. An intelligent adaptive wavefront measurement system for optical characterization, the system can obtain the optical parameters of a sample to be measured, the system comprises a light source, a shear generator and a photoelectric coupling device arranged in sequence along the axial direction, wherein: The shear generator includes a spatial light modulator and a phase difference microlens array which are sequentially arranged along the axial direction.
2. The system according to claim 1, wherein: The phase difference microlens array comprises front convex microlenses and rear convex microlenses arranged alternately on adjacent grids, and the phase difference microlens array performs π phase modulation on the incident light wave through the height difference between the top of the front convex microlens and the top of the rear convex microlens in the axial direction.
3. The system according to claim 1, wherein: In the subsequent wavefront reconstruction operation, the light transmittance of the top of the front convex microlens is calculated as 1, the light transmittance of the top of the base of the phase difference microlens array is calculated as 0, and the light transmittance of the top of the rear convex microlens is calculated as -1, wherein the height difference is Wherein, n is the refractive index of the material of the phase difference microlens array, and λ is the incident wavelength.
4. The system according to claim 1, wherein: The duty ratio of each microlens in the grid of the phase difference microlens array is: Among them, r is the aperture radius of each microlens, and d is the grid spacing.
5. The system according to claim 4, wherein: The duty cycle is 2:
3.
6. The system according to claim 1, wherein: The shear ratio of the shear generator is Among them, s is the shear amount, D is the diameter of the incident light, and L is the distance between the photoelectric coupling device and the phase difference microlens array, H is the thickness of the substrate of the phase difference microlens array, λ is the incident wavelength, and d is the grid spacing.
7. The system according to claim 1, wherein: The spatial light modulator modulates a shearing parameter, and the shearing parameter includes at least one of the following: a duty cycle of each microlens of the phase difference microlens array in its grid, a shearing amount, and an illumination intensity.
8. The system according to claim 1, wherein: The spatial light modulator optimizes the modulation mode of the spatial light modulator based on a machine learning model, so that the machine learning model can output a modulation mode that is adaptive to the sample to be tested according to the spectrum signal under the initial modulation mode.
9. The system according to claim 8, wherein: The machine learning model is established using a deep convolutional neural network.
10. The system according to claim 8, wherein: The machine learning model can optimize the modulation mode of the spatial light modulator so that the spectrum signal corresponding to the wavefront light intensity distribution received on the optoelectronic coupling device only has 0th order and ±1st order.
11. The system according to claim 9, wherein: The training process of the machine learning model can be carried out in a simulation environment. By designing a sample to be tested with a complex structure, the optical parameters of the designed structure are obtained in the simulation. Since the optical parameters are known, the optical parameters are used as the true value of the deep convolutional neural network training, and through continuous iterative learning and training, a machine learning model that has completed the training is finally obtained.
12. An intelligent adaptive wavefront measurement method for optical characterization, the method comprising the following steps: a) an incident light wave from a light source passes through a sample to be tested, a shear generator and a photoelectric coupling device arranged in sequence along the axial direction to obtain a wavefront light intensity distribution, wherein the shear generator includes a spatial light modulator and a phase difference microlens array arranged in sequence along the axial direction; b) obtaining a frequency spectrum signal corresponding to the wavefront light intensity distribution after Fourier transformation of the wavefront light intensity distribution; c) performing inverse Fourier transform and shearing interference signal processing on the spectrum signal to obtain the optical parameters of the sample to be tested.
13. The method according to claim 12, wherein: The phase difference microlens array comprises front convex microlenses and rear convex microlenses arranged alternately on adjacent grids, and the phase difference microlens array performs π phase modulation on the incident light wave through the height difference between the top of the front convex microlens and the top of the rear convex microlens in the axial direction.
14. The method according to claim 12, wherein: The light transmittance of the front convex microlens is 1, the light transmittance of the bottom part is 0, and the light transmittance of the rear convex microlens is -1, wherein the height difference is Wherein, n is the refractive index of the material of the phase difference microlens array, and λ is the incident wavelength.
15. The method according to claim 12, wherein: The duty ratio of each microlens in the grid of the phase difference microlens array is: Among them, r is the aperture radius of each microlens, and d is the grid spacing.
16. The method according to claim 15, wherein: The duty cycle is 2:
3.
17. The method according to claim 12, wherein: The shear ratio of the shear generator is Among them, s is the shear amount, D is the diameter of the incident light, and L is the distance between the photoelectric coupling device and the phase difference microlens array, H is the thickness of the substrate of the phase difference microlens array, λ is the incident wavelength, and d is the grid spacing.
18. The method according to claim 12, wherein: The spatial light modulator modulates a shearing parameter, wherein the shearing parameter comprises at least one of the following: a duty cycle of each microlens of the phase difference microlens array in its grid, a shearing amount and an illumination intensity.
19. The method according to claim 12, wherein: The spatial light modulator optimizes the modulation mode of the spatial light modulator based on a machine learning model, so that the machine learning model can output a modulation mode that is adaptive to the sample to be tested according to the spectrum signal under the initial modulation mode.
20. The method according to claim 19, wherein: The machine learning model is established using a deep convolutional neural network.
21. The method according to claim 19, wherein: The machine learning model can optimize the modulation mode of the spatial light modulator so that the spectrum signal corresponding to the wavefront light intensity distribution received on the optoelectronic coupling device only has 0th order and ±1st order.
22. The method according to claim 20, wherein: The training process of the machine learning model can be carried out in a simulation environment. By designing a sample to be tested with a complex structure, the optical parameters of the designed structure are obtained in the simulation. Since the optical parameters are known, the optical parameters are used as the true value of the deep convolutional neural network training, and through continuous iterative learning and training, a machine learning model that has completed the training is finally obtained.
23. A training method for an intelligent adaptive wavefront measurement method for optical characterization in a simulation environment, the method comprising the following steps: a) in a simulation environment, an incident light wave from a light source passes through a sample with known optical parameters, a spatial light modulator, a phase difference microlens array and a photoelectric coupling device arranged in sequence along the axial direction to obtain an initial wavefront light intensity distribution, wherein the spatial light modulator works in an initial modulation mode; b) obtaining an initial frequency spectrum signal corresponding to the initial wavefront light intensity distribution after Fourier transformation of the initial wavefront light intensity distribution; c) inputting the initial spectrum signal and the current modulation mode into a machine learning model established using a deep convolutional neural network, extracting and reconstructing features of the input signal through deep convolution, and outputting an adjusted modulation mode; d) Calculate the remaining level of the spectrum signal at this time and include it in the training loss for model training; e) obtaining an adjusted wavefront light intensity distribution based on the adjusted modulation mode, and obtaining an adjusted frequency spectrum signal corresponding to the adjusted wavefront light intensity distribution after Fourier transformation of the adjusted wavefront light intensity distribution; f) performing inverse Fourier transform and shearing interferometer signal processing on the adjusted spectrum signal to obtain the optical parameters of the sample, and calculating the error between the obtained optical parameters and the known true values of the optical parameters, and accounting for the training loss; g) Iteratively train the model according to the obtained training loss, and finally obtain a machine learning model that has completed the training.
24. An intelligent adaptive wavefront measurement method for optical characterization, the method comprising the following steps: a) an incident light wave from a light source passes through a sample to be tested, a spatial light modulator, a phase difference microlens array and a photoelectric coupling device arranged in sequence along the axial direction to obtain an initial wavefront light intensity distribution, wherein the spatial light modulator works in an initial modulation mode; b) obtaining an initial frequency spectrum signal corresponding to the initial wavefront light intensity distribution after Fourier transformation of the initial wavefront light intensity distribution; c) inputting the initial spectrum signal and the current modulation mode into the trained machine learning model, Extract and reconstruct the features of the input signal through deep convolution, and adjust the modulation mode so that the spectrum signal corresponding to the wavefront light intensity distribution received on the photoelectric coupling device only has 0th order and ±1st order; d) adjusting the modulation mode of the spatial light modulator based on the adjusted modulation mode to obtain an optimized wavefront light intensity distribution, and obtaining an optimized spectrum signal corresponding to the optimized wavefront light intensity distribution after Fourier transformation of the optimized wavefront light intensity distribution; e) performing inverse Fourier transform and shearing interference signal processing on the optimized spectrum signal to obtain the optical parameters of the sample to be tested.
Citation Information
Patent Citations
Digital phase-shifting lateral shearing interferometer and optical system wave aberration measurement method
CN104807548A
Angular displacement detecting method for adaptive optics system, imaging magnification detecting method for adaptive optics system, and adaptive optics system
CN105264346A
Machine learning fast aberration measuring system and method based on wave-front sensor
CN110346340A
In-situ surface shape splicing measurement device and method for large-aperture planar optical element
CN112577446A
Interference quantitative phase microscopic imaging system capable of being used for real-time detection of living cells
CN114965365A
Cited By
Automatic detection system for in-vehicle electronic streaming media rearview mirror
CN120669040A
