Optical element surface defect depth detection method and system based on iterative phase retrieval algorithm

Abstract

The invention discloses an optical element surface defect depth detection method and system based on an iterative phase retrieval algorithm, and relates to the field of optical element surface defect depth detection.The system comprises an imaging module and a control module, the imaging module is used for obtaining a focusing intensity image and a defocusing intensity image of a sample; and the control module is used for realizing defect depth detection according to the focusing intensity image and the defocusing intensity image of the sample by adopting an optical element surface defect depth detection method based on an iterative phase recovery algorithm. According to the invention, an accurate depth detection result can be rapidly obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection and phase recovery, and in particular to a method and system for detecting the depth of surface defects of optical elements based on an iterative phase recovery algorithm. Background Art

[0002] Amidst the booming modern optical industry, the application of optical components has expanded to encompass numerous key areas. The optical systems comprised of these components play a crucial role in cutting-edge optical and military fields, including high-power lasers, large-scale integrated circuits, and medical devices. For example, optical devices such as low-light-level night vision devices and infrared night vision devices convert and enhance images of scenery visible to the human eye at night or in other low-light conditions, providing technical support for military nighttime operations. Precision instruments rely on the precise machining and rigorous testing of their internal optical components. However, minute surface defects such as scratches and pitting on optical components can cause serious problems: they increase system noise, reduce image contrast, and ultimately affect the stability and reliability of the entire optical system. Therefore, developing more accurate and efficient methods for detecting surface defects in optical components has become a critical and pressing issue for the optical industry. This is not only crucial for technological advancement but also an essential requirement for promoting high-quality development in the industry.

[0003] The methods currently used to detect surface defects of optical components mainly include interference method, non-interference method and scattering method. Interference method has limitations in applicability due to its complex equipment and high experimental cost, so this detection method is generally not considered. For example, the Chinese invention patent with announcement number CN110006924B discloses a method for detecting the two-dimensional contour of tiny defects on the surface of optical components, which overcomes the problems of complex structure, low measurement accuracy, large data volume and long measurement time in the existing technology. However, since the image sensor can generally only collect the intensity information of the defect, and the phase information that characterizes the depth of the defect cannot be obtained, the method only realizes the detection of the two-dimensional information of the length and width of the defect, and fails to detect the important depth information of the defect. Therefore, it is very important to obtain the lost defect phase information through the phase recovery method.

[0004] Currently, two typical non-interferometric phase retrieval methods are based on the intensity transfer equation and on iteration. The intensity transfer equation method relates the intensity and phase information of the wave on the object plane to the rate of change of intensity along the propagation direction, converting the unknown phase variation into a visible intensity variation, thereby resolving the phase information on the object plane. This method has the advantages of being non-interferometric, non-iterative, and easy to implement. Another type of iterative phase retrieval algorithm is the GS algorithm proposed by Gerchberg and Saxton in 1972. Its basic idea is to use the known light field intensity to alternately map between the spatial domain and the Fourier domain, repeatedly iterating to constrain the object plane phase convergence. In recent years, with the research and development of phase retrieval algorithms, many new iterative methods and algorithms have emerged. These new algorithms improve on the original methods from different perspectives, improving the accuracy and convergence speed of phase retrieval, and bringing great vitality to the development of phase retrieval technology. In particular, the angular spectrum iterative phase retrieval algorithm overcomes the limitation of iterating between only two planes, expanding the possibilities of iteration, thereby reducing experimental constraints and improving the efficiency and accuracy of iteration. Since the algorithm does not involve the approximation of propagation distance in the processing process, it effectively reduces the experimental error and thus improves the accuracy of phase recovery.

[0005] For example, patent application CN115112016A discloses a method for three-dimensional surface defect detection on optical components based on angular spectrum iteration. This method overcomes the existing challenges of large data volumes, long measurement times, and low accuracy. However, this method's use of phase information to characterize defect depth requires consideration of the wavelength of light used in the experiment, and the angular spectrum iteration algorithm employed suffers from slow convergence.

[0006] For example, the invention patent application with publication number CN118348668A discloses a phase recovery method based on phase perturbation and gradient compensation. Although this method can obtain accurate defocus distance through electronically controlled focusing, it still requires manual movement of the sample to align the light source with the area to be tested of the sample, and there is interference from the initial random phase, which is not suitable for defect detection. Summary of the Invention

[0007] In order to solve the technical problems existing in the background technology, the present invention proposes a method and system for detecting the depth of surface defects of optical elements based on an iterative phase retrieval algorithm.

[0008] In a first aspect, the present invention proposes a method for detecting depth of surface defects of optical components based on an iterative phase retrieval algorithm, comprising:

[0009] S1, obtaining a focused intensity image and a defocused intensity image of the sample;

[0010] S2. Obtaining the phase of the sample according to the focused intensity image and the defocused intensity image of the sample;

[0011] S3, performing zero-filling processing on the focused intensity image, the defocused intensity image, and the phase of the sample, using the phase after the zero-filling processing as the initial phase, and combining the initial phase with the amplitude of the focused intensity image after the zero-filling processing to form a complex amplitude of the initial estimated focal plane of the sample;

[0012] S4. Determine the complex amplitude of the current focusing plane based on the initially estimated complex amplitude of the focusing plane; and use the amplitude of the zero-filled focus intensity image to constrain the complex amplitude of the current focusing plane to obtain an updated complex amplitude of the focusing plane;

[0013] S5. Perform forward angular spectrum propagation on the updated complex amplitude of the focused plane to the defocused plane to obtain the complex amplitude of the defocused plane. Simultaneously, add a linear weighting factor to the complex amplitude of the defocused plane. Use the amplitudes of the zero-filled focused intensity image and the defocused intensity image to perform frequency domain amplitude optimization weighting on the complex amplitude of the defocused plane to obtain the updated complex amplitude of the defocused plane.

[0014] S6. Perform reverse angular spectrum propagation on the updated complex amplitude of the defocused plane to propagate it to the focused plane to regain the complex amplitude of the focused plane, perform gradient descent processing on the phase of the regained complex amplitude of the focused plane to obtain a gradient-compensated phase; use the gradient-compensated phase as the phase for the next iteration, and replace the phase of the complex amplitude of the focused plane with the gradient-compensated phase to obtain the complex amplitude of the focused plane for the next iteration;

[0015] S7, determine whether the preset number of iterations has been reached; if not, proceed to S8; if so, proceed to S9;

[0016] S8, taking the complex amplitude of the focus plane of the next iteration as the complex amplitude of the focus plane of the initial estimation, and proceeding to S4;

[0017] S9, ending the iteration, taking the phase after gradient compensation as the final phase information, and using the phase modulation characteristic to convert the final phase information into depth information, thereby completing the depth detection of the defect.

[0018] Preferably, in S2, obtaining the phase of the sample according to the focused intensity image and the defocused intensity image of the sample specifically includes:

[0019] S21, performing differential processing on the focused intensity image and the defocused intensity image, giving an intensity transfer equation, and setting an original phase, intensity derivative, threshold, initial step size scaling factor, initial error value, and convergence condition;

[0020] S22, setting the intensity distribution on the focusing plane to a constant, simplifying the intensity transfer equation, and using fast Fourier transform to solve the equation to obtain an initial phase estimate;

[0021] S23, using the initial phase estimate to calculate the intensity derivative and the derivative difference, and obtaining a new phase estimate by using the gradient direction, the acceleration coefficient, the set original phase, and the initial phase estimate;

[0022] S24, using the new phase estimate and the set original phase to calculate a new error value and a new compensation scaling factor;

[0023] S25, judging whether the preset convergence condition is met based on the new phase estimate; if so, proceeding to S27; if not, proceeding to S26;

[0024] S26, using the new phase estimate as the original phase for the next iteration, and proceeding to S22;

[0025] S27. End the iteration and use the new phase estimate as the phase of the sample.

[0026] Preferably, the threshold E1=0.1, the threshold E2=0.0001, the initial step scaling factor P0=1, the initial error value R0=0, the original phase ψ0=0, the intensity derivative Let the derivative difference of the intensity derivative be ΔD0, where ΔD0=D0; the convergence condition is that the total number of iterations N is reached.

[0027] Preferably, the initial phase estimate is

[0028]

[0029] Where, I m =I(x,y,z) represents the intensity distribution on the focal plane and is a constant; represents the phase estimate of the focal plane returned by the n-1th iteration, ΔD n-1 represents the derivative difference at the n-1th iteration, n = 1, 2, …, N.

[0030] Preferably, the intensity derivative is

[0031]

[0032] Where D n represents the intensity derivative at the nth iteration, and k represents the wave number.

[0033] Preferably, the derivative difference is

[0034] ΔD n =D n -D n-1 ;

[0035] Where ΔD n Denotes the derivative difference at the nth iteration, D n represents the intensity derivative at the nth iteration, D n-1 represents the intensity derivative at the n-1th iteration.

[0036] Preferably, the new phase estimate is

[0037]

[0038] Where g n represents the gradient direction at the nth iteration, β n represents the acceleration coefficient at the nth iteration, ψ n represents the phase estimate of the new focal plane obtained at the nth iteration, ψ n-1 represents the phase estimate of the new focal plane obtained at the n-1th iteration, P n-1 Indicates the step size scaling factor at the n-1th iteration, represents the phase estimate of the focus plane returned by iteration n-1.

[0039] Preferably, the new error value is

[0040]

[0041] Where R n is the error value of the nth iteration.

[0042] Preferably, the new step size scaling factor is

[0043]

[0044] Where R n-1 is the error value of the n-1th iteration, P n Indicates the step size scaling factor returned at the nth iteration.

[0045] Preferably, the calculation formula for depth information is:

[0046]

[0047] Where h is the depth information, ψ q+1 is the final phase information, λ is the wavelength of light, q=1, 2,…, Q, and Q is the preset number of iterations.

[0048] In a second aspect, the present invention further proposes a system for detecting depth of surface defects of optical elements based on an iterative phase retrieval algorithm, comprising:

[0049] An imaging module, used for acquiring a focused intensity image and a defocused intensity image of the sample;

[0050] A control module is configured to implement defect depth detection based on a focused intensity image and a defocused intensity image of a sample using the optical element surface defect depth detection method based on an iterative phase recovery algorithm as described in any one of the first aspects.

[0051] Preferably, the imaging module includes a base, a bracket, a variable light source, a collimating lens, a stage, an objective lens, a tube lens, a CCD camera and a lifting drive mechanism;

[0052] The variable light source is fixed above the base; the collimating lens is arranged directly above the variable light source and fixedly connected to the variable light source; the bracket is fixed to the base on one side of the variable light source, the stage is slidably connected to the bracket in the vertical direction, and the light beam hole of the stage is located directly above the collimating lens; the objective lens is located directly above the light beam hole of the stage, and the objective lens is fixedly connected to the bracket, the barrel lens is located directly above the objective lens, and the barrel lens is fixedly connected to the bracket; the CCD camera is located directly above the barrel lens, and the CCD camera is fixedly connected to the bracket; the lifting drive mechanism is connected to the stage, and the lifting drive mechanism is used to drive the stage to slide in the vertical direction.

[0053] Preferably, the stage is further provided with a positioning assembly, the positioning assembly comprising a first electric control rod, a second electric control rod, a transverse positioning plate and a longitudinal positioning plate, the transverse positioning plate being arranged on the stage along the width direction of the stage, the longitudinal positioning plate being arranged on the stage along the length direction of the stage, and the tops of the transverse positioning plate and the longitudinal positioning plate both having a support surface for supporting the sample;

[0054] The first electric control rod is fixed on the loading platform, and the free end of the first electric control rod is fixedly connected to the transverse positioning plate. The first electric control rod is used to drive the transverse positioning plate to move longitudinally. The second electric control rod is fixed on the loading platform, and the free end of the second electric control rod is fixedly connected to the longitudinal positioning plate. The second electric control rod is used to drive the longitudinal positioning plate to move transversely.

[0055] Preferably, the stage is further provided with a clamping assembly for fixing the sample located on the positioning assembly.

[0056] Preferably, the control module includes:

[0057] Light source module, used to control the variable light source to switch the light source color and brightness;

[0058] The electric control module is used to control the movement of the lifting drive mechanism, the first electric control rod and the second electric control rod to achieve control of the axial displacement distance of the stage and horizontal movement of the sample to be tested;

[0059] An image processing module, configured to detect defect depth based on a focused intensity image and a defocused intensity image of the sample;

[0060] Central control module, used to control the light source module, electronic control module and image processing module;

[0061] Preferably, the control module further includes:

[0062] The mobile device is used to receive and display the focused intensity image and the defocused intensity image of the sample and the defect depth detection result.

[0063] In the present invention, the proposed method and system for depth detection of surface defects of optical elements based on an iterative phase recovery algorithm uses an iterative phase recovery algorithm that integrates an improved intensity transfer equation, and utilizes focused intensity images and defocused intensity images to achieve depth detection of defects. Specifically, an improved intensity transfer equation solution method is used to preliminarily solve the phase, accelerate the convergence speed of the angular spectrum iteration algorithm, improve the accuracy of the sample phase recovery results of the angular spectrum iteration algorithm, and obtain more accurate depth detection results; a step size scaling factor is added when solving the intensity transfer equation to avoid the step size shrinking too quickly, which causes the error value to converge to a non-minimum point; and adding optimized weighting and gradient descent during the iteration process can prevent the algorithm from falling into a local minimum, accelerate the convergence speed of the algorithm, and thus quickly obtain accurate depth detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a flow chart of a method for detecting depth of surface defects of optical elements based on an iterative phase retrieval algorithm in one embodiment of the present invention.

[0065] Figure 2 Schematic diagram of the optical path of an optical element surface defect depth detection system based on an iterative phase retrieval algorithm in one embodiment of the present invention.

[0066] Figure 3 This is a schematic structural diagram of an optical element surface defect depth detection system based on an iterative phase retrieval algorithm in one embodiment of the present invention.

[0067] Figure 4 Schematic diagram of the structure of the loading platform in one embodiment of the present invention.

[0068] Figure 5 These are the focus intensity diagram and defocus intensity diagram in one embodiment of the present invention; wherein, (a1) is the focus intensity diagram of a fine scratch, (a2) is the defocus intensity diagram of a fine scratch with a defocus distance of 40 μm, (b1) is the focus intensity diagram of a coarse scratch, and (b2) is the defocus intensity diagram of a coarse scratch with a defocus distance of 40 μm.

[0069] Figure 6The figures are the result images obtained by using different algorithms for defect depth detection; among them, (a1) is the three-dimensional distribution map of fine scratches reconstructed by the angular spectrum iteration algorithm, (a2) is the longitudinal distribution map of fine scratches reconstructed by the ASI algorithm, (a3) ​​is the three-dimensional distribution map of coarse scratches reconstructed by the ASI algorithm, (a4) is the longitudinal distribution map of coarse scratches reconstructed by the ASI algorithm, (b1) is the three-dimensional distribution map of fine scratches reconstructed by the adaptive angular spectrum iteration (AASI) algorithm, (b2) is the longitudinal distribution map of fine scratches reconstructed by the AASI algorithm, (b3) is the three-dimensional distribution map of coarse scratches reconstructed by the AASI algorithm, (b4) is the longitudinal distribution map of coarse scratches reconstructed by the AASI algorithm, (c1) is the three-dimensional distribution map of fine scratches reconstructed by the iterative phase retrieval (HIPR) algorithm, (c2) is the longitudinal distribution map of fine scratches reconstructed by the HIPR algorithm, (c3) is the three-dimensional distribution map of coarse scratches reconstructed by the HIPR algorithm, and (c4) is the longitudinal distribution map of coarse scratches reconstructed by the HIPR algorithm.

[0070] Figure 7 The root mean square error curves obtained by using different algorithms for defect depth detection; among them, (a1) is the root mean square error curve of fine scratches reconstructed by different algorithms, and (a2) is the RMSE curve of coarse scratches reconstructed by different algorithms. DETAILED DESCRIPTION

[0071] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0072] First, refer to Figure 1 The present invention proposes a method for detecting depth of optical element surface defects based on an iterative phase retrieval algorithm, comprising:

[0073] S1, obtaining a focused intensity image and a defocused intensity image of the sample;

[0074] S2. Obtaining the phase of the sample according to the focused intensity image and the defocused intensity image of the sample;

[0075] S3, performing zero-filling processing on the focused intensity image, the defocused intensity image, and the phase of the sample, using the phase after the zero-filling processing as the initial phase, and combining the initial phase with the amplitude of the focused intensity image after the zero-filling processing to form a complex amplitude of the initial estimated focal plane of the sample;

[0076] S4. Determine the complex amplitude of the current focusing plane based on the initially estimated complex amplitude of the focusing plane; and use the amplitude of the zero-filled focus intensity image to constrain the complex amplitude of the current focusing plane to obtain an updated complex amplitude of the focusing plane;

[0077] S5. Perform forward angular spectrum propagation on the updated complex amplitude of the focused plane to the defocused plane to obtain the complex amplitude of the defocused plane. Simultaneously, add a linear weighting factor to the complex amplitude of the defocused plane. Use the amplitudes of the zero-filled focused intensity image and the defocused intensity image to perform frequency domain amplitude optimization weighting on the complex amplitude of the defocused plane to obtain the updated complex amplitude of the defocused plane.

[0078] S6. Perform reverse angular spectrum propagation on the updated complex amplitude of the defocused plane to propagate it to the focused plane to regain the complex amplitude of the focused plane, perform gradient descent processing on the phase of the regained complex amplitude of the focused plane to obtain a gradient-compensated phase; use the gradient-compensated phase as the phase for the next iteration, and replace the phase of the complex amplitude of the focused plane with the gradient-compensated phase to obtain the complex amplitude of the focused plane for the next iteration;

[0079] S7, determine whether the preset number of iterations has been reached; if not, proceed to S8; if so, proceed to S9;

[0080] S8, taking the complex amplitude of the focus plane of the next iteration as the complex amplitude of the focus plane of the initial estimation, and proceeding to S4;

[0081] S9, ending the iteration, taking the phase after gradient compensation as the final phase information, and using the phase modulation characteristic to convert the final phase information into depth information, thereby completing the depth detection of the defect.

[0082] The present invention uses an iterative phase recovery algorithm that integrates an improved intensity transfer equation and utilizes focused intensity images and defocused intensity images to achieve depth detection of defects. The improved intensity transfer equation solution method is used to initially solve the phase, accelerating the convergence rate of the angular spectrum iteration algorithm and improving the accuracy of the sample phase recovery results of the angular spectrum iteration algorithm, thereby obtaining more accurate depth detection results. A step size scaling factor is added when solving the intensity transfer equation to prevent the step size from shrinking too quickly, causing the error value to converge to a non-minimum point. Adding optimized weighting and gradient descent during the iteration process can prevent the algorithm from falling into a local minimum, accelerating the convergence rate of the algorithm, and thus quickly obtaining accurate depth detection results.

[0083] The sample in this embodiment is an optical element.

[0084] In this embodiment, in S2, the phase of the sample is obtained according to the focused intensity image and the defocused intensity image of the sample, which specifically includes:

[0085] S21, performing differential processing on the focused intensity image and the defocused intensity image, giving an intensity transfer equation, and setting an original phase, intensity derivative, threshold, initial step size scaling factor, initial error value, and convergence condition;

[0086] S22, setting the intensity distribution on the focusing plane to a constant, simplifying the intensity transfer equation, and using fast Fourier transform to solve the equation to obtain an initial phase estimate;

[0087] S23, using the initial phase estimate to calculate the intensity derivative and the derivative difference, and obtaining a new phase estimate by using the gradient direction, the acceleration coefficient, the set original phase, and the initial phase estimate;

[0088] S24, using the new phase estimate and the set original phase to calculate a new error value and a new compensation scaling factor;

[0089] S25, judging whether the preset convergence condition is met based on the new phase estimate; if so, proceeding to S27; if not, proceeding to S26;

[0090] S26, using the new phase estimate as the original phase for the next iteration, and proceeding to S22;

[0091] S27. End the iteration and use the new phase estimate as the phase of the sample.

[0092] This embodiment adds a step size scaling factor when solving the intensity transfer equation to avoid the error value converging to a non-minimum point due to excessive step size contraction, thereby effectively improving the accuracy of the phase of the subsequently obtained samples.

[0093] The intensity transfer equation in this embodiment is:

[0094]

[0095] Where I(x,y,z) and represent the light field intensity distribution and phase estimation of the focused intensity image, respectively, represents the gradient operator, k represents the wave number, is called the intensity derivative.

[0096] The preset conditions in this embodiment are:

[0097] Threshold E1 = 0.1, threshold E2 = 0.0001, initial step scaling factor P0 = 1, initial error value R0 = 0, original phase ψ0 = 0, intensity derivative Let the derivative difference of the intensity derivative be ΔD0, where ΔD0=D0; the convergence condition is that the total number of iterations N is reached.

[0098] in,

[0099] Where, I m =I(x,y,z) represents the intensity distribution on the focal plane and is a constant; represents the phase estimate of the focal plane returned by the n-1th iteration, ΔDn-1 represents the derivative difference of the n-1th iteration, k represents the wave number, and n = 1, 2, …, N.

[0100] in, Where D n represents the intensity derivative of the nth iteration, k represents the wave number; ΔD n =D n -D n-1 Where, ΔD n represents the derivative difference at the nth iteration.

[0101] in,

[0102]

[0103] Where g n represents the gradient direction at the nth iteration, β n represents the acceleration coefficient at the nth iteration, ψ n represents the phase estimate of the new focal plane obtained at the nth iteration, ψ n-1 represents the phase estimate of the new focal plane obtained at the n-1th iteration, P n-1 Indicates the step size scaling factor at the n-1th iteration, represents the phase estimate of the focus plane returned by iteration n-1.

[0104] The new error value is Where R n is the error value of the nth iteration; preferably, the new step size scaling factor is

[0105]

[0106] Where R n-1 is the error value of the n-1th iteration, P n Indicates the step size scaling factor returned at the nth iteration.

[0107] In this embodiment, U1=|U|exp(jψ′); where U1 represents the complex amplitude of the initially estimated focal plane, |U| represents the amplitude of the focus intensity image obtained after zero-filling processing; ψ′ represents the initial phase obtained after zero-filling processing; exp represents an exponential function with base e, and j represents an imaginary unit.

[0108] In this embodiment, U q =|U q |exp(jψ q ); where q represents the number of iterations, q=1,2,...,Q, Q is the preset number of iterations, U qrepresents the complex amplitude of the focal plane obtained at the qth iteration, |U q | represents the amplitude of the complex amplitude of the focal plane obtained in the qth iteration, ψ q represents the phase of the complex amplitude of the focal plane obtained in the qth iteration.

[0109] In this embodiment, Where, Represents the complex amplitude of the updated focus plane.

[0110] In this embodiment, the calculation formula of the complex amplitude of the defocused surface is U q′ =|U q′ |exp(jθ q );

[0111] Where U q′ represents the complex amplitude of the defocused surface obtained at the qth iteration, |U q′ | represents the amplitude of the complex amplitude of the defocused surface obtained in the qth iteration, θ q represents the phase of the complex amplitude of the defocused surface obtained at the qth iteration.

[0112] In the linear weighting process, ω=exp(|U|-|U q′ |-ln2); where ω represents the linear weighting factor, which represents the weight of the amplitude of the complex amplitude of the defocused surface;

[0113] Among them, |G q |=2|U|-ω|U q′ |; In the formula, |G q | is the amplitude obtained after amplitude optimization weighting at the qth iteration;

[0114] Among them, U q″ =|G q |exp(jθ q );where U q″ is the complex amplitude of the defocused surface obtained after linear weighting processing at the qth iteration.

[0115] In this embodiment, in S7, the updated complex amplitude of the defocused plane is subjected to reverse angular spectrum propagation to the focused plane to obtain the complex amplitude of the focused plane again, and the phase of the complex amplitude of the recovered focused plane is subjected to gradient descent processing to obtain the complex amplitude of the focused plane of the next iteration, which specifically includes:

[0116] The updated complex amplitude U of the defocused surface q′ Perform reverse angular spectrum propagation to the focusing plane and regain the complex amplitude of the focusing plane; where, Where, represents the complex amplitude of the defocused surface after propagating through the reverse angular spectrum to the complex amplitude of the focused surface; |U q+1| represents the amplitude of the complex amplitude of the defocused surface after propagating through the reverse angular spectrum to the focused surface, It represents the phase of the complex amplitude of the defocused surface after propagating through the reverse angular spectrum to the focused surface;

[0117] Phase of the complex amplitude at the focal plane Perform gradient descent processing, and the phase of the complex amplitude at the focal plane Add an iterative additional value α q h q , and obtain the phase ψ after gradient compensation q+1 ; and the phase ψ after gradient compensation q+1 as the phase for the next iteration;

[0118] The phase of the complex amplitude of the focal plane The phase ψ after gradient compensation q+1 Perform substitution to obtain the complex amplitude of the focus plane for the next iteration.

[0119] in, Where, t q represents the phase difference of the complex amplitude of the focal plane in the qth iteration, ψ q represents the phase of the complex amplitude of the focal plane obtained at the qth iteration, It represents the phase of the complex amplitude of the defocused surface after propagating through the reverse angular spectrum to the focused surface.

[0120] in, Where, α q Indicates the acceleration factor.

[0121] in, Where, represents the phase of the complex amplitude of the defocused surface after propagation through the reverse angular spectrum to the focused surface in the qth iteration, It represents the phase of the complex amplitude of the defocused surface after propagation through the reverse angular spectrum to the focused surface in the q-1th iteration.

[0122] in, Where h q represents the gradient direction, which is proportional to the phase difference between the current iteration and the most recent iteration, ψ q+1 is the phase after gradient compensation.

[0123] Among them, U q+1 =|U q+1 |exp(jψ q+1 ); where |U q+1 | is the amplitude of the complex amplitude of the focal plane for the next iteration, ψ q+1 is the phase after gradient compensation, that is, the phase of the complex amplitude of the focal plane for the next iteration.

[0124] In this embodiment, the calculation formula of depth information is: Where h is the depth information, ψ q+1 is the final phase information, and λ is the wavelength of light.

[0125] This embodiment is configured in this way to accurately obtain defect depth information.

[0126] Second, as Figure 2-Figure 4 As shown, the present invention also proposes an optical element surface defect depth detection system based on an iterative phase recovery algorithm, including: an imaging module and a control module:

[0127] The imaging module is used to obtain a focused intensity image and a defocused intensity image of the sample;

[0128] The control module is used to implement defect depth detection based on the focused intensity image and the defocused intensity image of the sample by using the optical element surface defect depth detection method based on the iterative phase recovery algorithm described in any one of the first aspects.

[0129] The control module in this embodiment uses the optical element surface defect depth detection method based on the iterative phase recovery algorithm described in any one of the first aspects to perform defect depth detection on the focus intensity image and defocus intensity image of the sample obtained by the imaging module, and can quickly obtain accurate depth detection results.

[0130] In this embodiment, the imaging module includes a variable light source 1, a collimating lens 2, a stage 3, an objective lens 4, a tube lens 5, a CCD camera 6, a base 7 and a bracket 8; the stage is used to place an optical element sample; the variable light source 1 is used to emit a light beam of a preset wavelength; the collimating lens 2 is used to collimate the light beam into a first parallel light beam and then irradiate the light beam onto the optical element sample placed on the stage 3; the objective lens 4 is used to receive the light beam transmitted by the optical element; the tube lens 5 is used to collimate the light transmitted by the objective lens 4 into a second parallel light beam and irradiate the light onto the CCD camera; the CCD camera 6 is used to form an image based on the second parallel light beam.

[0131] This embodiment uses a variable light source 1 to meet the detection needs of different samples, and has wide applicability and high flexibility.

[0132] Specifically, the variable light source 1 is fixed above the base 7 by screws; the collimating lens 2 is arranged directly above the variable light source 1 and fixedly connected to the variable light source 1; the bracket 8 is fixed on the base 7 on one side of the variable light source 1, the stage 3 is fixed to the bracket 8 by screws, and the beam hole of the stage is located directly above the collimating lens 2; the objective lens 4 is located directly above the beam hole of the stage, and the objective lens 4 is fixedly connected to the bracket 8, the barrel lens 5 is located directly above the objective lens 4, and the barrel lens 5 is fixedly connected to the bracket 8; the CCD camera 6 is located directly above the barrel lens 5, and the CCD camera 6 is fixedly connected to the bracket 8.

[0133] During operation, the variable light source 1 emits a corresponding light beam according to the wavelength requirement; after being collimated into a parallel light beam by the collimating lens 2, it irradiates the sample placed on the stage 3; then, the light beam passes through the objective lens 4, is magnified by the objective lens 4, and is focused between the objective lens 4 and the tube lens 5, and passes through the tube lens 5 and is collimated by the tube lens 5. Finally, the light beam is directly imaged on the CCD camera 6.

[0134] In a further embodiment, the imaging module also includes a lifting drive mechanism, the stage 3 is connected to the bracket 8 in a sliding manner in the vertical direction, and the lifting drive mechanism is connected to the stage 3. The lifting drive mechanism is used to drive the stage 3 to slide in the vertical direction to facilitate electronic focus adjustment to obtain an accurate defocus distance.

[0135] Specifically, the lifting drive mechanism includes at least one set of drive components, each of which includes a motor 9 and a ball screw. The motor 9 and the ball screw are fixed coaxially, and the nut of the ball screw is fixedly connected to the stage 3. The ball screw is used to convert the rotational motion of the motor 9 into linear motion of the nut, which can drive the stage 3 up and down.

[0136] In a further embodiment, a positioning component for positioning the sample is further provided on the stage 3 , and the positioning component is applicable to optical elements of different specifications.

[0137] Specifically, the positioning assembly includes a first electric control rod 301, a second electric control rod 303, a transverse positioning plate 302, and a longitudinal positioning plate 303. The transverse positioning plate 302 is arranged on the stage 3 along the width direction of the stage 3, and the longitudinal positioning plate 303 is arranged on the stage 3 along the length direction of the stage 3. The tops of the transverse positioning plate 302 and the longitudinal positioning plate 303 each have a support surface for supporting the sample.

[0138] The first electric control rod 301 is fixed on the loading platform 3, and the free end of the first electric control rod 301 is fixedly connected to the transverse positioning plate 302. The first electric control rod 301 is used to drive the transverse positioning plate 302 to move longitudinally. The second electric control rod 303 is fixed on the loading platform 3, and the free end of the second electric control rod 303 is fixedly connected to the longitudinal positioning plate 303. The second electric control rod 303 is used to drive the longitudinal positioning plate 303 to move transversely.

[0139] In this embodiment, the control module sets the defocus distance and drives the motor 9, the first electric control rod 301, and the second electric control rod 303 to accurately move the stage 3 and the sample to be tested, thereby avoiding the accumulated errors caused by the need to move the sample to be tested multiple times during the traditional acquisition process, so that the system has higher accuracy and can perform quantitative analysis.

[0140] In a further embodiment, a clamping assembly for fixing the sample is further provided on the stage 3 .

[0141] Specifically, the clamping assembly includes at least one rotating pressure plate 305 for rotational fixation, which can cooperate with the first electric control rod 301 and the second electric control rod 303 to fix the sample, so that the sample remains stable when the stage 3 moves axially (i.e., in the vertical direction), thereby avoiding damage to the sample due to movement of the stage 3.

[0142] Of course, in other embodiments, other existing clamping and fixing structures may also be used to fix the samples on the transverse positioning plate 302 and the longitudinal positioning plate 303 .

[0143] In a further embodiment, the control module includes: a light source module, an electronic control module, an image processing module, and a central control module; the light source module is electrically connected to the variable light source 1, the electronic control module is electrically connected to the lifting drive mechanism, the first electronic control rod 301, and the second electronic control rod 303 respectively, the image processing module is electrically connected to the forming module, and the light source module, the electronic control module and the image processing module are electrically connected to the central control module respectively.

[0144] Among them, the light source module is used to control the variable light source 1 to achieve changes in the color and brightness of the light source; the electric control module is used to control the movement of the lifting drive mechanism, the first electric control rod 301, and the second electric control rod 303 to achieve control of the axial displacement distance of the stage 3 and horizontal movement of the sample to be tested; the image processing module is used to perform defect depth detection based on the focus intensity image and defocus intensity image of the sample; the central control module is used to regulate the light source module, the electric control module and the image processing module.

[0145] In a further embodiment, the control module also includes a mobile device, which is electrically connected to the central control module and the image processing module to receive and display the focus intensity image and the defocus intensity image, and the defect depth detection results, so that the user can view the images taken by the system and the results of the defect detection.

[0146] In this embodiment, after the CCD camera 6 captures the intensity image of the sample required for defect detection, the system's image processing module can process the intensity image and perform defect detection, and can display the defect detection results of the sample on a mobile device or computer, with high real-time and interactive capabilities.

[0147] In one specific embodiment, when using the optical element surface defect depth detection system based on the iterative phase retrieval algorithm proposed by the present invention to detect scratch defects, the specific operation process includes:

[0148] S1. Adjust the system light source: call the light source module in the central control module through the mobile device to turn on the variable light source 1 or control the brightness and color of the variable light source 1.

[0149] S2. Adjust the system status and shoot the focus intensity image of the sample: first, call the electric control module in the central control module through the mobile device, control the first electric control rod 301 and the second electric control rod 303 to drive the optical element sample to move horizontally on the stage 3, and find the most suitable observation position; then, fix the optical element sample through the clamping assembly; then, call the electric control module through the mobile device to control the motor 9 to drive the stage 3 to move in the axial direction of the optical path, and find the most suitable focusing position, so that the image on the CCD camera 6 is focused; finally, the central control module will call the CCD camera 6 to shoot an image, and the image obtained by shooting is the focus intensity image of the sample, and the image is sent to the image processing module for temporary storage and processing.

[0150] S3. Capturing a defocus intensity image of the sample: The electronic control module is called in the central control module via a mobile device. The defocus distance is input, and the motor 9 drives the stage 3 to move precisely the corresponding distance in the axial direction of the optical path. After the stage 3 reaches the corresponding position, the main control module automatically calls the CCD camera 6 to capture an image. The captured image is the defocus intensity image of the sample, which is then sent to the image processing module for temporary storage and processing.

[0151] S4. Defect detection: The image processing module is called in the central control module through a mobile device. In this module, all the captured focus intensity images and defocus intensity images can be viewed, and defect detection is performed on these focus intensity images and defocus intensity images using an optical element surface defect depth detection method based on an iterative phase recovery algorithm proposed in the present invention.

[0152] S5. Viewing and sharing of results: After defect detection is completed, the test results of the optical component sample will be displayed on the mobile device, and the results can also be transferred to the computer via a USB data cable for viewing.

[0153] The optical element sample used in this embodiment is a cylindrical acrylic scratch plate as the experimental sample. The experimental parameters are as follows: a cylindrical acrylic scratch plate with a thickness of 5mm and a diameter of 20mm, with fine scratches of 0.1um depth and 10um width and coarse scratches of 0.1um depth and 10um width engraved on the surface; the magnification of the microscope is selected to be 10 times, and the defocus distance is 40um. The focus intensity image and defocus intensity image of the optical element sample are shown as follows: Figure 5 shown.

[0154] The experimental results of defect detection using the optical element defect detection method based on iterative phase retrieval algorithm described in the present invention are as follows: Figure 6 and Figure 7 shown.

[0155] The root mean square error is usually expressed as RMSE, and the formula is defined as follows:

[0156]

[0157] Where h′ is the scratch depth detected by the algorithm, h is the original scratch depth, and M and N are the sizes of the scratch focus intensity image. The smaller the RMSE, the closer the restored image is to the original image.

[0158] In the experiment, the original measured depth values ​​of fine and coarse scratches were 0.1 μm. The ASI algorithm calculated the depth values ​​of 0.0915 μm for fine and 0.0869 μm for coarse scratches, with root mean square errors of 5.54% and 13.14% respectively. The AASI algorithm calculated the depth values ​​of 0.0928 μm for fine and 0.0895 μm for coarse scratches, with root mean square errors of 4.97% and 6.47% respectively. The HIPR algorithm calculated the depth values ​​of 0.0959 μm for fine and 0.0957 μm for coarse scratches, with root mean square errors of 4.11% and 4.31% respectively. It can be seen that the scratch depth calculated by the method of the present invention (i.e., the HIPR algorithm) is closer to the original measured value, indicating that the method of the present invention has better performance in defect detection.

[0159] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for detecting the depth of optical element surface defects based on an iterative phase retrieval algorithm, characterized in that: include: S1, obtaining a focused intensity image and a defocused intensity image of the sample; S2. Obtaining the phase of the sample according to the focused intensity image and the defocused intensity image of the sample; S3, performing zero-filling processing on the focused intensity image, the defocused intensity image, and the phase of the sample, using the phase after the zero-filling processing as the initial phase, and combining the initial phase with the amplitude of the focused intensity image after the zero-filling processing to form a complex amplitude of the initial estimated focal plane of the sample; S4. Determine the complex amplitude of the current focusing plane based on the initially estimated complex amplitude of the focusing plane; The amplitude of the zero-filled focus intensity image is used to constrain the complex amplitude of the current focus plane to obtain the updated complex amplitude of the focus plane. S5. Perform forward angular spectrum propagation on the updated complex amplitude of the focused plane to the defocused plane to obtain the complex amplitude of the defocused plane. Simultaneously, add a linear weighting factor to the complex amplitude of the defocused plane. Use the amplitudes of the zero-filled focused intensity image and the defocused intensity image to perform frequency domain amplitude optimization weighting on the complex amplitude of the defocused plane to obtain the updated complex amplitude of the defocused plane. S6. Perform reverse angular spectrum propagation on the updated complex amplitude of the defocused plane to propagate it to the focused plane to regain the complex amplitude of the focused plane, perform gradient descent processing on the phase of the regained complex amplitude of the focused plane to obtain a gradient-compensated phase; use the gradient-compensated phase as the phase for the next iteration, and replace the phase of the complex amplitude of the focused plane with the gradient-compensated phase to obtain the complex amplitude of the focused plane for the next iteration; S7, determine whether the preset number of iterations has been reached; if not, proceed to S8; if so, proceed to S9; S8, taking the complex amplitude of the focus plane of the next iteration as the complex amplitude of the focus plane of the initial estimation, and proceeding to S4; S9, ending the iteration, taking the phase after gradient compensation as the final phase information, and using the phase modulation characteristic to convert the final phase information into depth information, thereby completing the depth detection of the defect.

2. The method for detecting depth of optical element surface defects based on iterative phase retrieval algorithm according to claim 1, characterized in that: In S2, the phase of the sample is obtained according to the focused intensity image and the defocused intensity image of the sample, specifically including: S21, performing differential processing on the focused intensity image and the defocused intensity image, giving an intensity transfer equation, and setting an original phase, intensity derivative, threshold, initial step size scaling factor, initial error value, and convergence condition; S22, setting the intensity distribution on the focusing plane to a constant, simplifying the intensity transfer equation, and using fast Fourier transform to solve the equation to obtain an initial phase estimate; S23, using the initial phase estimate to calculate the intensity derivative and the derivative difference, and obtaining a new phase estimate by using the gradient direction, the acceleration coefficient, the original phase, and the initial phase estimate; S24, using the new phase estimate and the set original phase to calculate a new error value and a new compensation scaling factor; S25, judging whether the preset convergence condition is met based on the new phase estimate; if so, proceeding to S27; if not, proceeding to S26; S26, using the new phase estimate as the original phase for the next iteration, and proceeding to S22; S27. End the iteration and use the new phase estimate as the phase of the sample.

3. The method for detecting depth of optical element surface defects based on iterative phase retrieval algorithm according to claim 2, wherein: The intensity transfer equation is Where I(x,y,z) and represent the light field intensity distribution and phase estimation of the focused intensity image, respectively, represents the gradient operator, k represents the wave number, is the intensity derivative; Among them, threshold E1 = 0.1, threshold E2 = 0.0001, initial step scaling factor P0 = 1, initial error value R0 = 0, original phase ψ0 = 0, intensity derivative Let the derivative difference of the intensity derivative be ΔD0, where ΔD0 = D0; the convergence condition is to reach the total number of iterations N; Preferably, the initial phase estimate is Where, I m = I(x,y,z) represents the intensity distribution on the focal plane; represents the phase estimate of the focal plane returned by the n-1th iteration, ΔD n-1 represents the derivative difference of the n-1th iteration, k represents the wave number, n = 1, 2, ..., N; Preferably, the intensity derivative is Where D n represents the intensity derivative at the nth iteration; Preferably, the derivative difference is ΔD n =D n -D n-1 ; Where ΔD n Denotes the derivative difference at the nth iteration, D n represents the intensity derivative at the nth iteration, D n-1 represents the intensity derivative at the n-1th iteration; Preferably, the new phase estimate is Where g n represents the gradient direction at the nth iteration, β n represents the acceleration coefficient at the nth iteration, ψ n represents the phase estimate of the new focal plane obtained at the nth iteration, ψ n-1 represents the phase estimate of the new focal plane obtained at the n-1th iteration, P n-1 Indicates the step size scaling factor at the n-1th iteration, represents the phase estimate of the focal plane returned by the n-1th iteration; Preferably, the new error value is Where R n is the error value of the nth iteration; Preferably, the new step size scaling factor is Where R n-1 is the error value of the n-1th iteration, P n Indicates the step size scaling factor returned at the nth iteration.

4. The method for detecting depth of optical element surface defects based on iterative phase retrieval algorithm according to claim 1, wherein: The calculation formula for depth information is: Where h is the depth information, ψ q+1 is the final phase information, λ is the wavelength of light, q=1, 2,…, Q, and Q is the preset number of iterations.

5. An optical element surface defect depth detection system based on iterative phase retrieval algorithm, characterized in that: include: An imaging module, used for acquiring a focused intensity image and a defocused intensity image of the sample; A control module is used to implement defect depth detection based on the focused intensity image and the defocused intensity image of the sample using the optical element surface defect depth detection method based on the iterative phase recovery algorithm as described in any one of claims 1 to 4.

6. The optical element surface defect depth detection system based on iterative phase retrieval algorithm according to claim 5, characterized in that: The imaging module includes a base, a bracket, a variable light source, a collimating lens, a stage, an objective lens, a tube lens, a CCD camera and a lifting drive mechanism; The variable light source is fixed above the base; the collimating lens is arranged directly above the variable light source and fixedly connected to the variable light source; the bracket is fixed to the base on one side of the variable light source, the stage is slidably connected to the bracket along the vertical direction, and the beam hole of the stage is located directly above the collimating lens; The objective lens is located directly above the beam hole of the stage and is fixedly connected to the bracket. The barrel lens is located directly above the objective lens and is fixedly connected to the bracket. The CCD camera is located directly above the barrel lens and is fixedly connected to the bracket. The lifting drive mechanism is connected to the stage and is used to drive the stage to slide in the vertical direction.

7. The optical element surface defect depth detection system based on iterative phase retrieval algorithm according to claim 6, characterized in that: The stage is also provided with a positioning assembly, which includes a first electric control rod, a second electric control rod, a transverse positioning plate and a longitudinal positioning plate. The transverse positioning plate is arranged on the stage along the width direction of the stage, and the longitudinal positioning plate is arranged on the stage along the length direction of the stage. The tops of the transverse positioning plate and the longitudinal positioning plate each have a support surface for supporting the sample. The first electric control rod is fixed on the loading platform, and the free end of the first electric control rod is fixedly connected to the transverse positioning plate. The first electric control rod is used to drive the transverse positioning plate to move longitudinally. The second electric control rod is fixed on the loading platform, and the free end of the second electric control rod is fixedly connected to the longitudinal positioning plate. The second electric control rod is used to drive the longitudinal positioning plate to move transversely.

8. The optical element surface defect depth detection system based on iterative phase retrieval algorithm according to claim 7, characterized in that: The stage is also provided with a clamping component for fixing the sample located on the positioning component.

9. The optical element surface defect depth detection system based on iterative phase retrieval algorithm according to claim 7, characterized in that: The control module includes: Light source module, used to control the variable light source to switch the light source color and brightness; The electric control module is used to control the movement of the lifting drive mechanism, the first electric control rod and the second electric control rod to achieve control of the axial displacement distance of the stage and horizontal movement of the sample to be tested; An image processing module, configured to detect defect depth based on a focused intensity image and a defocused intensity image of the sample; The central control module is used to control the light source module, electronic control module and image processing module.

10. The optical element surface defect depth detection system based on iterative phase retrieval algorithm according to claim 9, characterized in that: The control module also includes: The mobile device is used to receive and display the focused intensity image and the defocused intensity image of the sample and the defect depth detection result.

Images