An optical element defect detection method, device and storage medium

Abstract

The application provides an optical element defect detection method and device and a storage medium, and relates to the technical field of optical element defect detection. The method comprises the following steps: acquiring initial fluorescence intensity data of a to-be-tested element, and performing denoising processing on the initial fluorescence intensity data to obtain first fluorescence intensity data; acquiring an instrument response function, and acquiring second fluorescence intensity data according to the instrument response function and the first fluorescence intensity data; inputting the second fluorescence intensity data into a fluorescence exponential decay model, adopting a nonlinear fitting method, performing algorithm optimization on the nonlinear fitting method, and acquiring a fluorescence intensity value; acquiring a fluorescence intensity image according to the fluorescence intensity value, and performing digital image processing on the fluorescence intensity image to extract defect information from the fluorescence intensity image. The application extracts defect information from an image through high-precision fluorescence intensity imaging, and realizes detection of surface defects of an optical element.

Description

Technical Field

[0001] The present invention relates to the technical field of optical element defect detection, and in particular to an optical element defect detection method, device and storage medium. Background Art

[0002] With the continuous development of advanced optical manufacturing technology, application fields such as microelectronics equipment, aerospace, precision measurement, and laser systems have put forward higher technical requirements for the surface processing quality of precision optical components, their core components. In particular, in some high-precision optical systems, there are strict control requirements for the surface defects of optical components.

[0003] Currently, fluorescence intensity is the most widely used method in optical imaging systems. For example, quantum dots are added to the polishing fluid during the polishing process of optical components as fluorescent markers, and laser confocal scanning is used to observe the structure and distribution of fluorescent defects in the optical components. Experiments have shown that the laser damage threshold of optical components is highly correlated with the fluorescence characteristics exhibited, which can be used to detect and evaluate the quality of components after machining. Other optical component defect detection systems based on fluorescence intensity have also been built. While the imaging principle of fluorescence intensity is relatively simple to implement, the measurement results are easily affected by environmental factors, making quantitative analysis difficult. Summary of the Invention

[0004] The problem to be solved by the present invention is how to achieve high-precision fluorescence intensity imaging, extract defect information from the image, and realize the detection of surface defects of optical components.

[0005] To solve the above problems, the present invention provides a method for detecting defects in an optical element, comprising:

[0006] Acquiring initial fluorescence intensity data of the device under test, and performing denoising processing on the initial fluorescence intensity data to obtain first fluorescence intensity data;

[0007] Obtaining an instrument response function, and obtaining second fluorescence intensity data based on the instrument response function and the first fluorescence intensity data;

[0008] Inputting the second fluorescence intensity data into a fluorescence exponential decay model, adopting a nonlinear fitting method, and performing algorithm optimization on the nonlinear fitting method to obtain a fluorescence intensity value;

[0009] A fluorescence intensity image is acquired according to the fluorescence intensity value, and digital image processing is performed on the fluorescence intensity image to extract defect information from the fluorescence intensity image.

[0010] Preferably, the obtaining of initial fluorescence intensity data of the device under test includes:

[0011] Acquire multiple sets of first data obtained by irradiating the laser onto different pixel areas on the surface of the device under test, each set of the first data including multiple sets of second data obtained by irradiating the laser onto the same pixel area at different times;

[0012] The performing denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data includes:

[0013] Obtaining the discrete degree of each group of the first data,

[0014] It is determined whether the first data is a noise signal according to the discrete degree. If so, the fluorescence intensity value of the pixel area corresponding to the first data is set to 0.

[0015] Preferably, obtaining the discreteness of each set of the first data includes:

[0016] Obtaining a standard deviation of each group of the first data, and obtaining a degree of dispersion of the first data according to the standard deviation;

[0017] The determining whether the first data is a noise signal according to the discreteness includes:

[0018] The standard deviation is compared with a set value, and when the standard deviation is smaller than the set value, the first data is determined to be a noise signal.

[0019] Preferably, the instrument response function is:

[0020]

[0021] Where G(t) is the instrument response function, F μ is the expectation of the instrument response function, F σ is the variance, t is the time;

[0022] The acquiring second fluorescence intensity data according to the instrument response function and the first fluorescence intensity data includes:

[0023] The fluorescence signal formed by the instrument response function is removed from the first fluorescence intensity data, and the moment when the fluorescence signal reaches a maximum value is used as the starting point of the first fluorescence intensity data.

[0024] Preferably, removing the fluorescence signal formed by the instrument response function from the first fluorescence intensity data comprises:

[0025] Obtaining the fluorescence intensity attenuation function according to the relationship between the first fluorescence intensity data and the instrument response function, and obtaining the second fluorescence intensity data according to the fluorescence intensity attenuation function;

[0026] The relationship between the first fluorescence intensity data and the instrument response function is as follows: the first fluorescence intensity data is the convolution of the fluorescence intensity attenuation function and the instrument response function.

[0027] Preferably, assuming that the second fluorescence intensity data is {(xi, yi), (i=1, 2, ..., m)}, the inputting the second fluorescence intensity data into the fluorescence exponential decay model and adopting the nonlinear fitting method include:

[0028] Estimating a function model with undetermined parameters, the function model is:

[0029] y = f(x);

[0030] Get the error distance of the i-th data, the error distance is:

[0031] f(xi)-yi;

[0032] Obtain the sum of the squares of the error distances of all data points S, which is:

[0033]

[0034] An objective function model corresponding to the minimum value of the sum of squares of error distances of all data points is obtained, where the objective function model is the fluorescence exponential decay model.

[0035] Preferably, the algorithm optimization of the nonlinear fitting method to obtain the fluorescence intensity value includes:

[0036] The parameters of the fluorescence exponential decay model are adjusted using the Levenberg-Marquardt algorithm to obtain a target fitting curve, and the fluorescence intensity value is obtained according to the target fitting curve.

[0037] Preferably, acquiring a fluorescence intensity image according to the fluorescence intensity value includes:

[0038] The fluorescence intensity value is converted into a fluorescence intensity image using a grayscale layering method, wherein the fluorescence intensity image is a pseudo-color image;

[0039] The performing digital image processing on the fluorescence intensity image to extract defect information from the fluorescence intensity image includes:

[0040] The fluorescence intensity image is subjected to denoising processing using a Gaussian low-pass filter to obtain a first image.

[0041] Performing grayscale stretching processing on the first image to obtain a second image, and performing sharpening and enhancement processing on the second image using a Laplace operator to obtain a third image,

[0042] Adaptively binarizing the third image to obtain a fourth image,

[0043] Performing smoothing and denoising processing on the fourth image to obtain a fifth image,

[0044] The missing pixel information in the fifth image is restored by a closing operation to obtain a sixth image.

[0045] A boundary tracing algorithm is used to extract the defect outline in the sixth image to determine the defect information.

[0046] The advantages of the optical element defect detection method of the present invention over the prior art are:

[0047] The optical component defect detection method of the present invention obtains initial fluorescence intensity data and performs denoising preprocessing to reduce calculation errors to obtain first fluorescence intensity data; then removes the fluorescence signal formed by the instrument response function in the first fluorescence intensity data to reduce the impact on the distribution of the final collected data, thereby obtaining second fluorescence intensity data; inputs the second fluorescence intensity data into a fluorescence exponential decay model, adopts a nonlinear fitting method, and optimizes the algorithm of the nonlinear fitting method to minimize the error between all sample data points and the estimated points of the fitting model, thereby obtaining a function model that best matches the original data; obtains fluorescence intensity values ​​based on the fitted optimized function model, and then obtains defect information based on the fluorescence intensity values ​​through image conversion and processing operations. The fitting algorithm adopted by the present invention is fast and has higher accuracy, capable of detecting very subtle surface defects of optical components.

[0048] The present invention also provides an optical element defect detection device, comprising:

[0049] a first acquisition unit, configured to acquire initial fluorescence intensity data of the device under test and perform denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data;

[0050] a second acquiring unit, configured to acquire an instrument response function and acquire second fluorescence intensity data according to the instrument response function and the first fluorescence intensity data;

[0051] a fitting optimization unit, configured to input the second fluorescence intensity data into a fluorescence exponential decay model, adopt a nonlinear fitting method, and perform algorithm optimization on the nonlinear fitting method to obtain a fluorescence intensity value;

[0052] A defect information extraction unit is used to obtain a fluorescence intensity image according to the fluorescence intensity value, perform digital image processing on the fluorescence intensity image, and extract defect information from the fluorescence intensity image.

[0053] The advantages of the optical element defect detection device and the optical element defect detection method described in the present invention are the same as those of the prior art, which will not be repeated here.

[0054] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is read and executed by a processor, the optical element defect detection method described in any one of the above items is implemented.

[0055] The advantages of the computer-readable storage medium of the present invention and the optical element defect detection method over the prior art are the same and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of the optical element defect detection method according to an embodiment of the present invention;

[0057] Figure 2 Schematic diagram of fluorescence signals at defective and non-defective locations on the surface of an optical element according to an embodiment of the present invention;

[0058] Figure 3 1 is a comparison diagram of the initial fluorescence intensity data before and after denoising in an embodiment of the present invention;

[0059] Figure 4 Schematic diagram of the relationship between initial fluorescence intensity data, fluorescence intensity decay curve and instrument response function in an embodiment of the present invention;

[0060] Figure 5 This is a comparison diagram of the first fluorescence intensity data before and after the instrument response function is removed in an embodiment of the present invention;

[0061] Figure 6 Flowchart of the Levenberg-Marquardt algorithm in an embodiment of the present invention;

[0062] Figure 7 Schematic diagram of an optical element defect detection device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0064] like Figure 1 As shown, an embodiment of the present invention provides a method for detecting defects in an optical element, comprising the following steps:

[0065] Step 110: acquiring initial fluorescence intensity data of the device under test, and performing denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data;

[0066] Step 120: obtaining an instrument response function, and obtaining second fluorescence intensity data based on the instrument response function and the first fluorescence intensity data;

[0067] Step 130: inputting the second fluorescence intensity data into a fluorescence exponential decay model, adopting a nonlinear fitting method, and performing algorithm optimization on the nonlinear fitting method to obtain a fluorescence intensity value;

[0068] Step 140: Acquire a fluorescence intensity image according to the fluorescence intensity value, perform digital image processing on the fluorescence intensity image, and extract defect information from the fluorescence intensity image.

[0069] To detect surface defects on optical components, this embodiment uses laser light to strike the surface of the optical component, obtaining reflection (scattering) information. This information is then used to infer the location and shape of the defects. Specifically, the laser strikes the component surface, generating fluorescence intensity data. This data is processed to select valid data, specifically data that reflects the defect. After extracting this data, it is further linked to the defect, thereby obtaining defect information.

[0070] This embodiment obtains first fluorescence intensity data by acquiring initial fluorescence intensity data and performing denoising preprocessing to reduce calculation errors. Second fluorescence intensity data is obtained by removing the fluorescence signal formed by the instrument response function from the first fluorescence intensity data to reduce its impact on the distribution of the final collected data. The second fluorescence intensity data is input into a fluorescence exponential decay model, and a nonlinear fitting method is employed. This nonlinear fitting method is algorithmically optimized to minimize the error between all sample data points and the estimated points of the fitting model, thereby obtaining a function model that best matches the original data. Fluorescence intensity values ​​are obtained based on the fitted, optimized function model, and defect information is then obtained based on the fluorescence intensity values ​​through image conversion and processing operations. The fitting algorithm employed in the present invention is fast and highly accurate, capable of detecting very subtle surface defects in optical components.

[0071] In some embodiments, obtaining initial fluorescence intensity data of the component to be tested includes:

[0072] A plurality of groups of first data are obtained when the laser is irradiated onto different pixel areas on the surface of the component to be tested, wherein each group of first data includes a plurality of second data obtained when the laser is irradiated onto the same pixel area at different times.

[0073] It is understood that the surface of the device under test is typically divided into different pixel regions. Laser irradiation of different pixel regions generates first data corresponding to the different pixel regions. The number of first data sets corresponds to the number of pixel regions. For the same pixel region, laser irradiation can be performed continuously or at different times, for example, once at regular intervals. This generates first data consisting of multiple second data sets, which serve as reflection or scattering information from the laser hitting the pixel region. Based on this information, the location and shape of the defects can be inferred.

[0074] In some embodiments, performing denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data includes:

[0075] Get the discrete degree of each group of first data,

[0076] It is determined whether the first data is a noise signal according to the discrete degree. If so, the fluorescence intensity value of the pixel area corresponding to the first data is set to 0.

[0077] In this embodiment, the denoising process for the initial fluorescence intensity data is mainly to eliminate data that may be noise signals based on the discreteness of the data.

[0078] For example, the laser is shot at different pixel areas on the surface of the optical element to obtain M groups of first data, each group of first data is the data obtained when the laser hits the surface of the optical element at different times, such as the number of photons. Different pixel areas may have defects or may not have defects, such as Figure 2 As shown, Figure 2 (a) is a set of fluorescence intensity data obtained at the surface defects of the component to be tested. Figure 2 (b) is a set of fluorescence intensity data obtained at the non-defective part of the surface of the component to be tested. It can be seen that there is a big difference in the distribution of the two sets of data. Figure 2 (a) The degree of dispersion of the data is much greater than Figure 2 (b) The messy noisy signal.

[0079] The discrete degree of each set of first data is used to determine whether it is a noise signal. If so, no subsequent data processing, i.e., calculation of the fluorescence intensity value, is performed. In an optional method, the fluorescence intensity value of the pixel area where the first data is obtained is directly set to 0.

[0080] In some implementation manners, obtaining the discreteness of each set of the first data includes:

[0081] Obtain the standard deviation of each group of the first data,

[0082] The dispersion degree of the first data is obtained according to the standard deviation.

[0083] Wherein, the standard deviation is expressed as:

[0084]

[0085] Among them, S represents the standard deviation, x i For each second data in each group of first data, is the mean of each second data in the first data.

[0086] This embodiment uses standard deviation as a measure of data distribution. Standard deviation reflects the degree of dispersion within a set of data. Due to its high sensitivity to deviation, the standard deviation not only reflects the differences reflected by the mean deviation, but also includes the differences between individual values ​​of the variable and the mean. This improves the reliability of the first fluorescence intensity data while removing background noise, effectively enhancing the quality of the resulting fluorescence intensity image.

[0087] This embodiment determines the degree of dispersion of the first data by obtaining the standard deviations of multiple groups of first data, thereby helping to remove noise signals in the first data.

[0088] In some implementations, determining whether the first data is a noise signal based on the degree of dispersion includes: comparing a standard deviation with a set value, and determining that the first data is a noise signal when the standard deviation is less than the set value.

[0089] This embodiment can improve the imaging quality of fluorescence intensity data after background denoising by using standard deviation to remove noise. Figure 3 As shown, Figure 3 (a) is the fluorescence intensity data of the defect without denoising. Figure 3 (b) is the fluorescence intensity data of the defect after denoising. It can be seen that Figure 3 (b) relatively Figure 3 (a) The imaging quality has been greatly improved.

[0090] In some embodiments, obtaining second fluorescence intensity data based on the instrument response function and the first fluorescence intensity data includes:

[0091] The fluorescence signal formed by the instrument response function is removed from the first fluorescence intensity data, and the moment when the fluorescence signal reaches a maximum value is used as the starting point of the first fluorescence intensity data.

[0092] The instrument response function formula is as follows:

[0093]

[0094] Where G(t) is the instrument response function, F μ is the expectation of the instrument response function, F σ is the variance, and t is the time.

[0095] While fluorescence decay data is a histogram with a discrete distribution along the time axis, it does not accurately reflect the fluorescence decay exponential model. The data acquisition system itself has a certain amount of system latency, which affects the time it takes the detector to detect photons and significantly influences the distribution of the resulting data. System latency stems from inherent factors such as the width of the excitation pulse, temporal scattering of the optical system, and timing jitter in the electronic system. These factors together form the system's temporal response function G(t) to the excitation pulse. This function is generally referred to as the instrument response function (IRF), which follows a Gaussian distribution and can be expressed as the above equation.

[0096] In some embodiments, removing the fluorescence signal formed by the instrument response function from the first fluorescence intensity data includes:

[0097] The fluorescence intensity decay function is obtained according to the relationship between the first fluorescence intensity data and the instrument response function,

[0098] Second fluorescence intensity data is obtained according to the fluorescence intensity decay function.

[0099] The relationship between the first fluorescence intensity data and the instrument response function is as follows: the first fluorescence intensity data is the convolution of the fluorescence intensity decay function and the instrument response function, which is expressed as:

[0100] F(t)=I(t)*G(t);

[0101] Wherein, F(t) is the first fluorescence intensity data, I(t) is the fluorescence intensity decay curve, and G(t) is the instrument response function.

[0102] In other words, the fluorescence decay data obtained in this example is actually a function derived from an exponential function after Gaussian correction. The IRF significantly affects the shape of the first half of the fluorescence exponential decay model, causing the effective fluorescence decay data to shift backward. Therefore, when analyzing the fluorescence decay data obtained in this example, the effect of the IRF on fluorescence intensity must be considered. To do so, this example removes the fluorescence signal generated by the IRF from the overall data and uses the moment when the fluorescence signal reaches its maximum value as the starting point of the data to obtain the fluorescence intensity decay data.

[0103] like Figure 4 As shown, Figure 4 Where F(t) represents the curve graph of the first fluorescence intensity data, which is obtained by the interaction of the fluorescence intensity decay function I(t) and the instrument response function G(t). The fluorescence intensity decay function is obtained by the deconvolution formula, the instrument response function is removed, and the moment when the fluorescence signal reaches the maximum value is used as the starting point of the first fluorescence intensity data. Figure 5 As shown, Figure 5 This is a set of comparison charts before and after the fluorescence intensity data is removed from the instrument response function. Figure 5 (a) is the data distribution diagram of the fluorescence intensity data collected by the system. Figure 5 (b) is the data distribution diagram of the fluorescence intensity data collected by the system without the instrument response function.

[0104] In some embodiments, assuming that the second fluorescence intensity data is {(xi,yi), (i=1,2,…m)}, the second fluorescence intensity data is input into the fluorescence exponential decay model, and the nonlinear fitting method is adopted, including:

[0105] The function model with unknown parameters is estimated as follows:

[0106] y = f(x);

[0107] Get the error distance of the i-th data, the error distance is:

[0108] f(xi)-yi;

[0109] Get the sum of the squares of the error distances of all data points S. The sum of the squares of the error distances of all data points is:

[0110]

[0111] The objective function model corresponding to the minimum value of the sum of squares of error distances of all data points is obtained, and the objective function model is a fluorescence exponential decay model.

[0112] The optical element defect detection method described in this embodiment uses a nonlinear fitting method to obtain a fluorescence exponential decay model, minimize the error between all sample data points and the fitting model estimation points, and obtain a function model that best matches the original data.

[0113] In some embodiments, the nonlinear fitting method is algorithmically optimized to obtain the fluorescence intensity value, including: adjusting the parameters of the fluorescence exponential decay model using the Levenberg-Marquardt algorithm to obtain a target fitting curve, and obtaining the fluorescence intensity value according to the target fitting curve.

[0114] like Figure 6 As shown, the parameters of the fluorescence exponential decay model are adjusted using the Levenberg-Marquardt algorithm, including:

[0115] Obtain the objective function x = f(p) and the initial point p0 of parameter estimation;

[0116] Substitute the initial point p0 into the formula ε0=||xf(p0)|| and calculate ε0, where k=0 and λ0=10 -3 , v = 10;

[0117] Calculate the Jacobi matrix J k , substitute into the formula Calculate N k , according to the N k , construct the incremental normal equation

[0118] Solving the incremental normal equation yields δ k , when ||xf(p k +δ k )||<ε k When p k+1 =p k +δ k , and ||δ k When ||<ε, stop the iteration and get the result, otherwise let λ k+1 =λ k / ν, recalculate the Jacobi matrix J k , repeat the above steps,

[0119] When ||xf(p k +δ k )||≥ε k When λ k+1 =λ k *ν, solving the incremental normal equation to obtain δ k , and re-obtain the initial point and repeat the above steps;

[0120] According to the objective function x = f(p), obtain the minimum point p of f(p) * .

[0121] The optical element defect detection method described in this embodiment uses the Levenberg-Marquardt algorithm for optimization and adjustment. The principle of the Levenberg-Marquardt algorithm is to approximate the calculation results of the parameter vector p that needs to be estimated in the fitting model within its neighborhood, ignore the second-order and above parts of the approximate terms, and convert the calculation process into a linear least squares problem. Therefore, the Levenberg-Marquardt algorithm is used to converge faster and the results obtained are more accurate.

[0122] In some embodiments, obtaining a fluorescence intensity image according to the fluorescence intensity value includes:

[0123] The fluorescence intensity value was converted into a fluorescence intensity image using the grayscale layering method, and the fluorescence intensity image was a pseudo-color image.

[0124] In some embodiments, performing digital image processing on the fluorescence intensity image to extract defect information from the fluorescence intensity image includes:

[0125] The fluorescence intensity image is subjected to denoising processing using a Gaussian low-pass filter to obtain a first image;

[0126] Performing grayscale stretching processing on the first image to obtain a second image, and performing sharpening and enhancement processing on the second image using a Laplace operator to obtain a third image;

[0127] Performing adaptive binarization processing on the third image to obtain a fourth image; performing smoothing and denoising processing on the fourth image to obtain a fifth image;

[0128] Restoring missing pixel information in the fifth image using a closing operation to obtain a sixth image;

[0129] A boundary tracing algorithm is used to extract the defect outline in the sixth image and determine the defect information.

[0130] The optical component defect detection method described in this embodiment uses a grayscale layering method to convert the fluorescence intensity values ​​into a fluorescence intensity image, thereby amplifying the differences in defect data at different locations. The fluorescence intensity image is then denoised using a Gaussian low-pass filter to improve its quality. The first image is then grayscale stretched. Defective regions generally have larger grayscale values. Grayscale stretching is performed at the high end of the grayscale value range to enhance the grayscale contrast of the image. The image is sharpened using a Laplacian operator to enhance areas with subtle defects. The resulting image is then adaptively binarized to enhance its visual quality. Denoising is then performed using the median filter to smooth the image, further improving its quality. Due to the filtering process, defect information is inevitably lost. Furthermore, due to the imaging system, the captured defect information is generally incomplete. These factors can cause a single defect pixel to be split into several smaller defects, hindering defect detection. A closing operation is then used to fill in smaller pixel gaps and discontinuities in the image through data processing, restoring the missing pixel information to a certain extent. Finally determine the defect information to make the obtained defect information more accurate.

[0131] Corresponding to the optical element defect detection method, the present invention also provides an optical element defect detection device, such as Figure 7 Shown, including:

[0132] A first acquiring unit 710 is configured to acquire initial fluorescence intensity data of the device under test and perform denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data;

[0133] A second acquiring unit 720, configured to acquire an instrument response function and acquire second fluorescence intensity data based on the instrument response function and the first fluorescence intensity data;

[0134] A fitting optimization unit 730 is configured to input the second fluorescence intensity data into a fluorescence exponential decay model, adopt a nonlinear fitting method, and perform algorithm optimization on the nonlinear fitting method to obtain a fluorescence intensity value;

[0135] The defect information extraction unit 740 is configured to obtain a fluorescence intensity image according to the fluorescence intensity value, perform digital image processing on the fluorescence intensity image, and extract defect information from the fluorescence intensity image.

[0136] The advantages of the optical element defect detection method device described in the present invention and the optical element defect detection method described above over the prior art are the same and will not be repeated here.

[0137] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is read and executed by a processor, the optical element defect detection method described in any one of the above items is implemented.

[0138] The advantages of the computer-readable storage medium of the present invention and the optical element defect detection method over the prior art are the same and will not be described in detail here.

[0139] Although the present disclosure is disclosed as above, the protection scope of the present disclosure is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present disclosure, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for detecting defects in an optical element, characterized in that: include: Acquiring initial fluorescence intensity data of the device under test, and performing denoising processing on the initial fluorescence intensity data to obtain first fluorescence intensity data; Obtaining an instrument response function, and obtaining second fluorescence intensity data based on the instrument response function and the first fluorescence intensity data; Inputting the second fluorescence intensity data into a fluorescence exponential decay model, adopting a nonlinear fitting method, and performing algorithm optimization on the nonlinear fitting method to obtain a fluorescence intensity value; acquiring a fluorescence intensity image according to the fluorescence intensity value, performing digital image processing on the fluorescence intensity image, and extracting defect information from the fluorescence intensity image; Wherein, the instrument response function is: ； Where G(t) is the instrument response function, is the expectation of the instrument response function, is the variance, t is the time; The acquiring second fluorescence intensity data according to the instrument response function and the first fluorescence intensity data includes: removing the fluorescence signal formed by the instrument response function from the first fluorescence intensity data, and taking the moment when the fluorescence signal reaches a maximum value as the starting point of the first fluorescence intensity data; Assume that the second fluorescence intensity data is , the second fluorescence intensity data is input into the fluorescence exponential decay model, and a nonlinear fitting method is adopted: Estimating a function model with undetermined parameters, the function model is: ； Get the The error distance of the data is: ； Obtain the sum of the squares of the error distances of all data points S, which is: ； An objective function model corresponding to the minimum value of the sum of squares of error distances of all data points is obtained, where the objective function model is the fluorescence exponential decay model.

2. The optical element defect detection method according to claim 1, wherein: The obtaining of initial fluorescence intensity data of the component to be tested comprises: Acquire multiple sets of first data obtained by irradiating the laser onto different pixel areas on the surface of the device under test, each set of the first data including multiple sets of second data obtained by irradiating the laser onto the same pixel area at different times; The performing denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data includes: Obtaining the discrete degree of each group of the first data, It is determined whether the first data is a noise signal according to the discrete degree. If so, the fluorescence intensity value of the pixel area corresponding to the first data is set to 0.

3. The optical element defect detection method according to claim 2, wherein: The obtaining of the discrete degree of each group of the first data includes: Obtaining a standard deviation of each group of the first data, and obtaining a degree of dispersion of the first data according to the standard deviation; The determining whether the first data is a noise signal according to the discreteness includes: The standard deviation is compared with a set value, and when the standard deviation is smaller than the set value, the first data is determined to be a noise signal.

4. The optical element defect detection method according to claim 1, wherein: The removing the fluorescence signal formed by the instrument response function from the first fluorescence intensity data comprises: Obtaining the fluorescence intensity attenuation function according to the relationship between the first fluorescence intensity data and the instrument response function, and obtaining the second fluorescence intensity data according to the fluorescence intensity attenuation function; The relationship between the first fluorescence intensity data and the instrument response function is as follows: the first fluorescence intensity data is the convolution of the fluorescence intensity attenuation function and the instrument response function.

5. The optical element defect detection method according to claim 1, wherein: Optimizing the algorithm of the nonlinear fitting method to obtain the fluorescence intensity value includes: The parameters of the fluorescence exponential decay model are adjusted using the Levenberg-Marquardt algorithm to obtain a target fitting curve, and the fluorescence intensity value is obtained according to the target fitting curve.

6. The optical element defect detection method according to claim 1, wherein: The acquiring of a fluorescence intensity image according to the fluorescence intensity value includes: The fluorescence intensity value is converted into a fluorescence intensity image using a grayscale layering method, wherein the fluorescence intensity image is a pseudo-color image; The performing digital image processing on the fluorescence intensity image to extract defect information from the fluorescence intensity image includes: The fluorescence intensity image is subjected to denoising processing using a Gaussian low-pass filter to obtain a first image. Performing grayscale stretching processing on the first image to obtain a second image, and performing sharpening and enhancement processing on the second image using a Laplace operator to obtain a third image, Adaptively binarizing the third image to obtain a fourth image, Performing smoothing and denoising processing on the fourth image to obtain a fifth image, The missing pixel information in the fifth image is restored by a closing operation to obtain a sixth image. A boundary tracing algorithm is used to extract the defect outline in the sixth image to determine the defect information.

7. An optical element defect detection device, characterized in that: The optical element defect detection method according to any one of claims 1 to 6 comprises: a first acquisition unit, configured to acquire initial fluorescence intensity data of the device under test and perform denoising on the initial fluorescence intensity data to obtain first fluorescence intensity data; a second acquiring unit, configured to acquire an instrument response function and acquire second fluorescence intensity data according to the instrument response function and the first fluorescence intensity data; a fitting optimization unit, configured to input the second fluorescence intensity data into a fluorescence exponential decay model, adopt a nonlinear fitting method, and perform algorithm optimization on the nonlinear fitting method to obtain a fluorescence intensity value; A defect information extraction unit is used to obtain a fluorescence intensity image according to the fluorescence intensity value, perform digital image processing on the fluorescence intensity image, and extract defect information from the fluorescence intensity image.

8. A computer-readable storage medium, characterized in that A computer program is stored, and when the computer program is executed by a processor, the optical element defect detection method according to any one of claims 1 to 6 is implemented.

Images