Iterative optimization detection device and method for complex surface topography of curved-surface micro-lens

Abstract

The invention relates to an iterative optimization detection device and method for complex surface topography of a curved-surface micro lens, which can effectively solve the problem of surface topography detection of a curved-surface micro lens array with both a substrate and a micro lens unit having high-gradient characteristics. The device comprises a laser (1), a convergent lens (2), a collimating lens (3), a deflection reflector (4), a spectroscope (5), a standard lens (6), a compensation lens (7), a to-be-measured curved surface microlens array (8), an inclination angle adjuster (9), an imaging lens (10), a filtering pinhole (11), ground glass (12), a camera (13) and a computer (14).

Description

Technical Field

[0001] This invention relates to the field of precision optics, and more particularly to an iterative optimization detection device for complex surface morphology of curved microlenses, and the method used in this iterative optimization detection device for complex surface morphology of curved microlenses. Background Technology

[0002] Curved microlens arrays, as a novel type of microlens array element, have developed rapidly in recent years, with patents such as CN201410806265.0 and CN202011132753.X reporting on this technology. Traditional microlens arrays are mainly planar microlens arrays, where the microlens units are distributed on a planar substrate, and the base height of each unit is consistent with the optical axis direction. In contrast, the microlens units of curved microlens arrays are distributed on a substrate, and the base height of each unit is not consistent with the optical axis direction.

[0003] The surface morphology of curved microlens arrays is essential for them to perform their intended functions, making surface morphology inspection crucial. Currently, the main inspection devices include atomic force microscopy, scanning electron microscopy, confocal microscopy, and white light interferometry. These technologies can achieve relatively high inspection accuracy and resolution; however, the first three require point-by-point scanning, which is inefficient. White light interferometry is suitable for inspecting large-area arrays, but for components like curved microlens arrays with large axial sag spans, it requires long-term, large-area axial scanning, also failing to achieve efficient inspection.

[0004] To address this issue, the applicant previously proposed a device and method for detecting complex surface morphology of curved microlens arrays (CN202311358828.X), which can detect and analyze the surface morphology of the substrate and microlens unit parts of the curved microlens array separately, exhibiting high flexibility. However, this method mainly compensates for substrates with high steepness, assuming that the steepness of the compensated microlens unit itself is small and will not have additional impact on the measurement. When the steepness of the microlens unit also increases synchronously, the following two problems arise: (1) the deviation between the wavefront of the detection beam and the surface morphology under test increases, causing the beam to fail to return to the system; (2) the hysteresis error increases, leading to a decrease in the calculation accuracy of the algorithm, and consequently a decrease in measurement accuracy.

[0005] The two problems mentioned above remain unresolved, and there is no more advantageous solution for the surface morphology detection of curved microlens arrays; further research is still needed. Summary of the Invention

[0006] To overcome the shortcomings of the prior art, the technical problem to be solved by the present invention is to provide an iterative optimization detection device for complex surface morphology of curved microlenses, which can effectively solve the problem of surface morphology detection of curved microlens arrays where both the substrate and the microlens unit have high steepness features.

[0007] The technical solution of the present invention is: This device for iterative optimization and detection of complex surface morphology of curved microlenses includes: a laser (1), a converging lens (2), a collimating lens (3), a deflecting mirror (4), a beam splitter (5), a standard mirror (6), a compensation lens (7), a microlens array of the surface to be tested (8), an tilt adjuster (9), an imaging lens (10), a filter pinhole (11), a frosted glass (12), a camera (13), and a computer (14);

[0008] A thin laser beam emitted from the laser is collimated into a wide-beam collimated laser beam after passing through a converging lens and a collimating lens. The laser beam then illuminates a deflecting mirror placed at 45° relative to the laser beam's optical axis. After reflection, the beam illuminates a beam splitter. The beam reflected by the beam splitter propagates forward and illuminates a standard mirror, the front surface of which serves as a reference surface. Part of the light illuminating the standard mirror returns along its original path after reflection from this surface, passing sequentially through the standard mirror, beam splitter, imaging lens, filter pinhole, and ground glass to reach the camera, serving as a reference beam. The other part of the light illuminating the standard mirror is transmitted through the standard mirror and continues to propagate forward. It then passes through a compensation lens and illuminates the surface of the microlens array to be measured. The beam illuminating the surface of the microlens array returns along its original path and then sequentially passes through the compensation lens, standard mirror, beam splitter, imaging lens, filter pinhole, and ground glass to reach the camera, serving as a measurement beam. The measurement beam and the reference beam interfere on the target surface of the camera, and the resulting interferogram is received by the camera and transmitted to the computer for calculation.

[0009] To address the problem of increased deviation between the wavefront of the detection beam and the morphology of the surface under test, which prevents the beam from returning to the system, the device uses a tilt angle adjuster to change the tilt angle of the microlens array of the surface under test relative to the system's optical axis. This alters the range of the beam that can return to the system, thus changing the detection range of the microlens array. The changed tilt angle is simultaneously recorded by a computer. After this adjustment, the optical path procedure described above is repeated to obtain the interferogram of the current detection range. This entire process is repeated until the detection range covers the entire area of ​​the microlens array. Solving all recorded interferograms and tilt angles yields the surface morphology of the microlens array.

[0010] A method for iterative optimization and detection of complex surface morphology of curved microlenses is also provided, which includes the following steps:

[0011] (1) Modeling the detection system: The backlash error is corrected by using the reverse iterative optimization method. The parameters of each component in the detection system, especially the standard lens and the compensation lens, are input into the computer, including the radius of curvature, focal length, thickness and spacing. Then, a virtual detection system model is built in the computer based on these parameters.

[0012] (2) Count the number of times the tilt angle is changed, record the interference pattern at each angle, and number them. Record each tilt angle as α. i Let i = 1, 2, ..., M, where M is the total number of times the tilt angle is changed, and the interferogram at each tilt angle is recorded as I. i (x,y), i=1,2,…,M, where x and y are the horizontal and vertical coordinates of the surface of the microlens array to be measured, respectively;

[0013] (3) Solve for the wavefront at the i-th tilt angle: For the interferogram I i The wavefront W is obtained by performing phase unwinding and phase unwrapping on (x,y). i (x,y);

[0014] (4) Divide the data range of the substrate and microlens units, use the edge extraction algorithm to extract the range of the microlens units, and then extract their phase data to distinguish them from the data within the substrate range. Record the wavefront within the substrate range as W. iBase (x,y), record the wavefront within the microlens unit as W. iMLA (x,y);

[0015] (5) Using Gaussian radial basis functions to analyze the wavefront W iMLA (x,y) is fitted and characterized. During fitting, the number of Gaussian radial basis functions used is not less than N. A Gaussian radial basis function is set at the center position of each microlens unit in step (4). The fundamental phase of these units is subtracted, and then the remaining part is fitted. Finally, the subtracted part is added back into the formula. At this time, the fitted wavefront is obtained, which is characterized as follows:

[0016]

[0017] Among them, W iMLA '(x,y) represents the wavefront fitting result, k is the index of the Gaussian radial basis function, K is the total number of Gaussian radial basis functions, and ε k w is the shape factor. k x is the coefficient. k and y k These are the center coordinates of the Gaussian radial basis functions;

[0018] (6) Evaluate the Gaussian radial basis function fitting residuals: Define the fitting residual evaluation threshold as ξ, and the relationship between the actual fitting residuals and this threshold satisfies:

[0019]

[0020] If this relationship cannot be satisfied, it means that the fitting result is incorrect, and the result should be substituted back into step (6) for calculation.

[0021] (7) Introduce the model of the microlens array of the surface to be measured into the detection system model. In step (1), the model of the microlens array of the surface to be measured is established in the detection system model. The surface morphology S of the microlens array of the surface to be measured is... iMLA '(x,y) is characterized by Gaussian radial basis functions:

[0022]

[0023] Where s k As a coefficient, the wavefront W of the detection system model is obtained by ray tracing based on the surface morphology of the microlens array under test and the detection system model. iMLA (x,y);

[0024] (8) Construct the objective function, and solve the surface morphology of the microlens of the surface under test at the current tilt angle through reverse iterative optimization. The coefficients s input in step (8) are... k Given an arbitrarily set initial value, the wavefront W obtained by solving the detection system model at this time... iMLA (x,y) and

[0025] W iMLA Since (x, y) are not equal, set the objective function as:

[0026] O([s1,s2,…,s K ])=min[(W iMLA (x,y)-W iMLA '(x,y)) 2 +c] (4) where min[] represents the minimum value condition, c is a constant term, and the coefficient s is continuously modified through iterative modification in the system. k This makes W iMLA (x,y) and W iMLA The surface morphology (x, y) is continuously approximated until it meets a sufficiently small requirement. At this point, it is assumed that the influence of the surface morphology of the microlens array of the test surface set in the system model on the system is approximately consistent with that of the actual surface morphology of the microlens array of the test surface. Therefore, this is taken as the actual surface morphology.

[0027]

[0028] Among them, S iMLA (x,y) represents the true surface morphology of the microlens unit portion of the microlens array under test at the i-th tilt angle;

[0029] (9) Switch the tilt angle, repeat the measurement, switch to the next tilt angle, and then repeat steps (3) to (8) until the surface morphology of the microlens unit part under all tilt angles is measured.

[0030] (10) Sub-aperture splicing is used to obtain the surface morphology of the microlens unit under test. The surface morphology of the spliced ​​microlens unit under test is denoted as S. MLA (x,y);

[0031] (11) Calculate the surface morphology error of the substrate. Since the device has already selected a compensating lens that matches the outgoing wavefront and the microlens substrate of the surface under test, the wavefront of the substrate should be consistent under different tilt angles. Therefore, the surface morphology of the substrate is calculated as follows:

[0032] S Base =W iBase (6);

[0033] (12) Calculate the complete surface morphology of the microlens on the surface to be measured, based on the substrate surface morphology error S in step (12). base Surface morphology of the microlens unit in step (11) And to compensate for the exit wavefront R of the lens, calculate the complete surface morphology S of the microlens array sample under test. Fin for:

[0034] Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the structure of the detection device for iterative optimization of complex surface morphology of curved microlenses according to the present invention.

[0036] Figure 2 The flowchart illustrates the iterative optimization detection method for complex surface morphology of curved microlenses according to the present invention. Detailed Implementation

[0037] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0038] It should be noted that the term "comprising" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products or devices.

[0039] like Figure 1 As shown, this device for iterative optimization and detection of complex surface morphology of curved microlenses includes: a laser 1, a converging lens 2, a collimating lens 3, a deflecting mirror 4, a beam splitter 5, a standard mirror 6, a compensation lens 7, a microlens array of the surface to be tested 8, a tilt adjuster 9, an imaging lens 10, a filter pinhole 11, a ground glass 12, a camera 13, and a computer 14.

[0040] A thin laser beam emitted from the laser is collimated into a wide-beam collimated laser beam after passing through a converging lens and a collimating lens. The laser beam then illuminates a deflecting mirror placed at 45° relative to the laser beam's optical axis. After reflection, the beam illuminates a beam splitter. The beam reflected by the beam splitter propagates forward and illuminates a standard mirror, the front surface of which serves as a reference surface. Part of the light illuminating the standard mirror returns along its original path after reflection from this surface, passing sequentially through the standard mirror, beam splitter, imaging lens, filter pinhole, and ground glass to reach the camera, serving as a reference beam. The other part of the light illuminating the standard mirror is transmitted through the standard mirror and continues to propagate forward. It then passes through a compensation lens and illuminates the surface of the microlens array to be measured. The beam illuminating the surface of the microlens array returns along its original path and then sequentially passes through the compensation lens, standard mirror, beam splitter, imaging lens, filter pinhole, and ground glass to reach the camera, serving as a measurement beam. The measurement beam and the reference beam interfere on the target surface of the camera, and the resulting interferogram is received by the camera and transmitted to the computer for calculation.

[0041] To address the problem of increased deviation between the wavefront of the detection beam and the morphology of the surface under test, which prevents the beam from returning to the system, the device uses a tilt angle adjuster to change the tilt angle of the microlens array of the surface under test relative to the system's optical axis. This alters the range of the beam that can return to the system, thus changing the detection range of the microlens array. The changed tilt angle is simultaneously recorded by a computer. After this adjustment, the optical path procedure described above is repeated to obtain the interferogram of the current detection range. This entire process is repeated until the detection range covers the entire area of ​​the microlens array. Solving all recorded interferograms and tilt angles yields the surface morphology of the microlens array.

[0042] Preferably, the center of the substrate of the microlens array of the surface under test coincides with the focal point of the compensation lens.

[0043] Preferably, the wavefront of the light emitted from the compensation lens is matched with the surface morphology of the substrate portion of the microlens array under test.

[0044] Preferably, the reference surface of the standard mirror and the surface of the microlens array of the surface under test are both uncoated glass surfaces. If the surface of the microlens array of the surface under test is a high-reflectivity surface or is coated with a high-reflectivity film, an attenuator is added between the standard mirror and the compensation lens to ensure the contrast of the interferogram.

[0045] Preferably, under the current parameters of the compensation lens and tilt adjuster, the detection range of the system covers the entire area of ​​the surface of the microlens array of the surface under test.

[0046] Preferably, the filter pinhole is selected based on the estimated density of the interference pattern fringe. The effective information in the interference pattern is in the low to mid-frequency range. The cutoff frequency of the pinhole is limited to 1 to 2 times the maximum density of the interference pattern to filter out noise.

[0047] like Figure 2 As shown, an iterative optimization detection method for complex surface morphology of curved microlenses is also provided, which includes the following steps:

[0048] (1) Modeling the detection system: The backlash error is corrected by using the reverse iterative optimization method. The parameters of each component in the detection system, especially the standard lens and the compensation lens, are input into the computer, including the radius of curvature, focal length, thickness and spacing. Then, a virtual detection system model is built in the computer based on these parameters.

[0049] (2) Count the number of times the tilt angle is changed, record the interference pattern at each angle, and number them. Record each tilt angle as α. i Let i = 1, 2, ..., M, where M is the total number of times the tilt angle is changed, and the interferogram at each tilt angle is recorded as I. i (x,y), i=1,2,…,M, where x and y are the horizontal and vertical coordinates (normal coordinates) of the surface of the microlens array to be measured, respectively;

[0050] (3) Solve for the wavefront at the i-th tilt angle: For the interferogram I i The wavefront W is obtained by performing phase unwinding and phase unwrapping on (x,y). i (x,y);

[0051] (4) Divide the data range of the substrate and microlens units, use the edge extraction algorithm to extract the range of the microlens units, and then extract their phase data to distinguish them from the data within the substrate range. Record the wavefront within the substrate range as W. iBase (x,y), record the wavefront within the microlens unit as W. iMLA (x,y);

[0052] (5) Using Gaussian radial basis functions to analyze the wavefront W iMLA (x,y) is fitted and characterized. During fitting, the number of Gaussian radial basis functions used is not less than N. A Gaussian radial basis function is set at the center position of each microlens unit in step (4). The fundamental phase of these units is subtracted, and then the remaining part is fitted. Finally, the subtracted part is refitted.

[0053] The newly added formula yields the fitted wavefront, which is characterized as follows:

[0054]

[0055] Among them, W iMLA '(x,y) represents the wavefront fitting result, k is the index of the Gaussian radial basis function, K is the total number of Gaussian radial basis functions, and ε k w is the shape factor. k x is the coefficient. k and y k These are the center coordinates of the Gaussian radial basis functions;

[0056] (6) Evaluate the Gaussian radial basis function fitting residuals: Define the fitting residual evaluation threshold as ξ, and...

[0057] The relationship between the actual fitting residual and the threshold satisfies:

[0058]

[0059] If this relationship cannot be satisfied, it indicates that the fitting result is incorrect, and the result should be resubmitted into step (6).

[0060] Perform calculations;

[0061] (7) Introduce the model of the microlens array of the surface to be measured into the detection system model. In step (1), the model of the microlens array of the surface to be measured is established in the detection system model. The surface morphology S of the microlens array of the surface to be measured is... iMLA '(x,y), through Gaussian radial basis functions

[0062] Line representation:

[0063]

[0064] Where s k As a coefficient, the wavefront of the detection system model is obtained by ray tracing based on the surface morphology of the microlens array under test and the detection system model.

[0065] W iMLA (x,y);

[0066] (8) Construct the objective function, and solve the surface morphology of the microlens of the surface under test at the current tilt angle through reverse iterative optimization. The coefficients s input in step (8) are... k Given an arbitrarily set initial value, the wavefront W obtained by solving the detection system model at this time... iMLA (x,y) and W iMLA Since (x, y) are not equal, set the objective function as:

[0067] O([s1,s2,…,s K ])=min[(W iMLA (x,y)-W iMLA '(x,y)) 2 +c] (4)

[0068] Where min[] represents the minimum value condition, and c is a constant term, which is obtained by iteratively modifying the coefficient s in the system. k This makes W iMLA (x,y) and W iMLA The surface morphology (x, y) is continuously approximated until it meets a sufficiently small requirement. At this point, it is assumed that the influence of the surface morphology of the microlens array of the test surface set in the system model on the system is approximately consistent with that of the actual surface morphology of the microlens array of the test surface. Therefore, this is taken as the actual surface morphology.

[0069]

[0070] Among them, S iMLA (x,y) represents the true surface morphology of the microlens unit portion of the microlens array under test at the i-th tilt angle;

[0071] (9) Switch the tilt angle, repeat the measurement, switch to the next tilt angle, and then repeat steps (3) to (8) until the microlens unit portion is measured at all tilt angles.

[0072] Surface morphology;

[0073] (10) Sub-aperture splicing is used to obtain the surface morphology of the microlens unit under test. The spliced ​​microlens unit is then used to obtain the surface morphology of the microlens unit under test.

[0074] The surface morphology of the micrometer lens unit is denoted as S. MLA (x,y);

[0075] (11) Calculate the surface morphology error of the substrate. Since the device has already selected a compensating lens that matches the outgoing wavefront and the microlens substrate of the surface under test, the wavefront of the substrate should be consistent under different tilt angles. Therefore, the surface morphology of the substrate is calculated as follows:

[0076] S Base =W iBase (6);

[0077] (12) Calculate the complete surface morphology of the microlens on the surface to be measured, based on the substrate surface morphology error S in step (12). Base Surface morphology of the microlens unit in step (11) And to compensate for the exit wavefront R of the lens, calculate the complete surface morphology S of the microlens array sample under test. Fin for:

[0078]

[0079] Preferably, when using Gaussian radial basis function fitting and inverse iterative optimization calculation, the total number of Gaussian radial basis functions K and the shape factor ε used are... k The center coordinates x of the Gaussian radial basis functions k and y k We must maintain consistency.

[0080] Preferably, in step (10), the sub-aperture stitching algorithm is a feature point matching method, an image cross-correlation algorithm, or a frequency domain transformation method. In actual use, the algorithm parameters can be adjusted according to engineering needs.

[0081] This invention addresses the need for surface topography detection of curved microlens arrays, specifically optimizing the solutions to the problems of beam return failure and large backhaul error mentioned in the background. In terms of the device, based on the principles of compensated interference and multi-angle stitching, the problem of beam return failure is solved through a compensating lens and tilt adjuster. In terms of the method, based on the principle of backhaul error correction, the problem of large backhaul error is solved through Gaussian radial basis functions and inverse iterative optimization calculations. Ultimately, this combination achieves precise detection of the surface topography of curved microlens arrays, with unique advantages in adaptability to high steepness.

[0082] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A device for detecting the complex surface profile of a curved microlens, characterized in that it comprises: It comprises: laser (1), converging lens (2), collimating mirror (3), deflection mirror (4), beam splitter (5), standard mirror (6), compensation lens (7), curved surface microlens array to be measured (8), tilt angle adjuster (9), imaging lens (10), filtering pinhole (11), ground glass (12), camera (13), computer (14); The fine laser beam emitted by the laser passes through the converging lens and the collimating mirror to form a wide-beam collimated laser beam, which is irradiated onto the deflection mirror placed at 45° relative to the optical axis of the laser beam, and is reflected by the deflection mirror to irradiate onto the beam splitter, wherein the light beam reflected by the beam splitter propagates forward and irradiates onto the standard mirror, the front surface of the standard mirror serving as a reference surface; part of the light irradiated onto the standard mirror is reflected by the surface to return to the original path, and sequentially passes through the standard mirror, the beam splitter, the imaging lens, the filtering pinhole, the ground glass to reach the camera, serving as a reference light beam; another part of the light irradiated onto the standard mirror is transmitted through the standard mirror and continues to propagate forward, and then passes through the compensation lens to irradiate onto the surface of the curved surface microlens array to be measured, and the light beam irradiated onto the surface of the curved surface microlens array to be measured returns to the original path, and then sequentially passes through the compensation lens, the standard mirror, the beam splitter, the imaging lens, the filtering pinhole, the ground glass to reach the camera, serving as a measurement light beam; the measurement light beam and the reference light beam interfere on the target surface of the camera to form an interference pattern, which is received by the camera and transmitted to the computer for calculation.

2. The device for detecting the complex surface profile of the curved microlens according to claim 1, wherein: The spherical center position of the base of the curved surface microlens array to be measured coincides with the focal point position of the compensation lens.

3. The device for detecting the complex surface profile of the curved microlens according to claim 2, wherein: The wavefront of the light wave emitted by the compensation lens matches the surface topography of the base part of the curved surface microlens array to be measured.

4. The device according to claim 3, wherein the device is characterized by: The reference surface of the standard mirror and the surface of the curved surface microlens array to be measured are both non-coated glass surfaces, and if the surface of the curved surface microlens array to be measured is a high-reflectivity surface or is coated with a high-reflectivity film, an attenuating sheet is added between the standard mirror and the compensation lens to ensure the contrast of the interference pattern.

5. The device for detecting the complex surface profile of the curved microlens according to claim 4, wherein: Under the current parameters of the compensation lens and the tilt angle adjuster, the detection range of the system covers the entire area of the surface of the curved surface microlens array to be measured.

6. The device for detecting the complex surface profile of the curved microlens according to claim 5, wherein: The filtering pinhole is selected under the condition of estimating the stripe density of the interference pattern, the effective information in the interference pattern is in the low and medium frequency, and the cutoff frequency of the pinhole is limited to 1-2 times the maximum density of the interference stripes to filter out noise.

7. A method for detecting the complex surface profile of a curved microlens, characterized in that: It comprises the following steps: (1) modeling the detection system: using the reverse iterative optimization method to correct the return error, inputting the parameters of each element in the detection system, especially the parameters of the standard lens and the compensation lens, including the curvature radius, the focal length, the thickness, the spacing, into the computer, and then establishing a virtual detection system model in the computer according to these parameters; (2) Count the number of times of changing the tilt angle, record the interference pattern at each angle, and number, for each tilt angle, record as a i , i = 1, 2, …, M, where M is the total number of times of changing the tilt angle, and the interference pattern at each tilt angle is recorded as I i (x, y), i = 1, 2, …, M, where x and y are the horizontal and vertical coordinates of the surface microlens array surface to be measured, respectively; (3) Calculate the wavefront at the ith tilt angle: phase unwrap and phase unwrap the interferogram I i (x,y) to obtain the wavefront W i (x,y); (4) Divide the base and microlens cell data range, adopt edge extraction algorithm to extract the range of microlens cell, and then extract its phase data to distinguish from the data in the base range, record the wavefront in the base range as W iBase (x,y), and record the wavefront in the microlens cell range as W iMLA (x,y). (5) using Gaussian radial basis functions to fit the wavefront W iMLA (x,y) to represent, in the fitting, the number of Gaussian radial basis functions used is not less than N, in step (4) each microlens unit center position is set a Gaussian radial basis function, the basic phase of these units is deducted, and then the remaining part is fitted, and finally the deducted part is added to the formula, and the fitted wavefront is obtained at this time, which is represented as: wherein W iMLA (x,y) is the fitting result of the wavefront, k is the serial number of the Gaussian radial basis function, K is the total number of the Gaussian radial basis functions, ε k is a shape factor, w k is a coefficient, x k and y k are the center coordinates of the Gaussian radial basis function, respectively. (6) evaluating the fitting residual of the Gaussian radial basis function: defining the fitting residual evaluation threshold as ξ, and the relationship between the actual fitting residual and the threshold satisfies: If this relationship cannot be satisfied, it means that the fitting result is incorrect, and the step (6) is repeated for operation; (7) Introducing the model of the to-be-tested curved microlens array into the detection system model, the model of the to-be-tested curved microlens array is introduced into the detection system model established in step (1), and the surface topography S iMLA '(x,y) is characterized by a Gaussian radial basis function: where s k is the coefficient, by the surface topography of the micro-lens array of the surface to be measured and the detection system model, the wave front of the detection system model at this time is solved by ray tracing as W iMLA (x, y); (8) constructing a target function, and solving the surface profile of the micro-lens of the measured surface under the current tilt angle by reverse iteration optimization, wherein the coefficient s k previously input in step (8) is an initial value set arbitrarily, at this time, the wave front W iMLA (x,y) and W iMLA '(x,y) solved in the detection system model are not equal, and the target function is set as: O([s1,s2,...,s K ]) = min[(W iMLA (x,y) - W iMLA '(x,y)) 2 +c] (4) where min[] denotes the minimum condition, c is a constant term, and the coefficient s is modified by iteration in the system k such that W iMLA '(x, y) is constantly approaching, until the requirement of being small enough is met, at which time the surface topography of the measured curved microlens array in the system model is considered to be approximately consistent with the real surface topography of the measured curved microlens array, and is further taken as the real surface topography iMLA '(x, y) is constantly approaching, until the requirement of being small enough is met, at which time the surface topography of the measured curved microlens array in the system model is considered to be approximately consistent with the real surface topography of the measured curved microlens array, and is further taken as the real surface topography S iMLA (x, y) is the real surface topography of the microlens unit part of the curved microlens array under the i-th tilt angle; (9) switching the tilt angle, repeating the measurement, switching to the next tilt angle, and then repeating steps (3) to (8) until the surface topography of the microlens unit part under all tilt angles is measured. (10) Sub-aperture stitching is used to obtain the surface topography of the micro-lens unit to be measured, and the surface topography of the micro-lens unit to be measured after stitching is denoted as S MLA (x,y); (11) Calculate the surface topography error of the substrate part. Since the device part has been selected with a compensation lens matching the exit wavefront and the surface of the micro-lens substrate to be measured, the wavefront of the substrate part should be consistent at different tilt angles. Therefore, the surface topography of the substrate part is calculated as: S Base = W iBase (6); (12) calculate the complete surface topography of the measured curved microlens, according to the substrate surface topography error S Base , the microlens unit surface topography in step (11) and the compensation lens exit wavefront R, calculate the complete surface topography S of the measured curved microlens array sample Fin :

8. The method of claim 7, wherein the method further comprises: The total number of Gaussian radial basis functions K, the shape factor ε k , the center coordinates x k and y k of the Gaussian radial basis functions must be kept consistent when using Gaussian radial basis function fitting and inverse iterative optimization calculation.

9. The method of claim 8, wherein: In the step (10), the sub-aperture stitching algorithm is a feature point matching method, an image cross-correlation method, or a frequency domain transformation method.

Images