A high-flux crystal material light intensity map acquisition method

Abstract

The present application relates to the technical field of light intensity atlas imaging, and particularly relates to a high-throughput crystal material light intensity atlas acquisition method, to solve the problems that the spatial resolution and atlas resolution are mutually limited in the prior art, and the small crystal grains and the limited crystal grain segmentation accuracy cannot be adapted to the problems. The present application acquires preset illumination conditions, uses a small numerical aperture optical imaging system to collect a low spatial resolution intensity image sequence, adopts an ultrahigh resolution wavefront reconstruction model to process a high spatial resolution complex amplitude field, and then performs grain boundary identification and grain segmentation, and finally forms a light intensity atlas by statistically analyzing the light intensity of the grains under different illumination conditions. The present application applies the ultrahigh resolution wavefront reconstruction technology to light intensity atlas acquisition, realizes high-throughput, high spatial resolution and high atlas resolution light intensity atlas acquisition, does not need additional hardware configuration, and provides a more efficient and accurate characterization method for crystal material research.

Description

Technical Field

[0001] This invention relates to the field of light intensity spectrum imaging technology, specifically to a high-throughput method for acquiring light intensity spectra of crystal materials. Background Technology

[0002] Crystal orientation imaging based on intensity maps is an emerging crystal orientation characterization technique. It has advantages such as high speed, large field of view, and no need for a vacuum environment, and has important value and potential in national defense security, information communication, and rail transportation.

[0003] Obtaining the light intensity spectrum of crystal materials is the core and key to crystal orientation optical imaging methods. The light intensity spectrum is used to describe the light intensity distribution of the crystal grains under different illumination conditions: when different angles of illumination are applied to the surface of the crystal material sample, or when reflected light is collected at different angles, the light intensity that the imaging system can collect will also change accordingly.

[0004] Existing methods for obtaining light intensity spectra of crystal materials have the following main problems:

[0005] First, spatial resolution and spectral resolution are mutually limiting; a light intensity spectrum cannot simultaneously possess both high spatial resolution and high spectral resolution. An optical imaging system is a low-pass filter system; a higher cutoff frequency means higher spatial resolution. This cutoff frequency is directly determined by the system's numerical aperture (NA). To achieve higher spatial resolution, an imaging system with a larger NA is required. A larger NA often necessitates a larger aperture angle for light collection, but this reduces spectral resolution because light from more directions can be collected, making changes in illumination conditions less influential on the light intensity in the imaging result.

[0006] Second, it cannot adapt to small grains: The basic principle of intensity pattern construction requires that the aperture angle of a single image cannot be too large, which means that existing methods can only use small NA objectives for imaging, thus limiting the spatial resolution of the optical imaging system to about 1 / 10 of its theoretical limit. Therefore, the size of the grains that can be detected is also limited, and it cannot adapt to small grains.

[0007] Third, grain segmentation accuracy is limited: Grain segmentation relies on grain boundary identification, which depends on the differences between grain boundaries and non-grain boundary regions in the image. However, these differences are sometimes small, especially at low image resolutions, which can easily lead to missed or false detections of grain boundaries, resulting in multiple different grains being segmented into a single grain. Because existing methods have limited information available for grain boundary identification and grain segmentation (low image resolution; only intensity images, no phase information), grain segmentation accuracy is limited. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides a high-throughput method for obtaining light intensity spectra of crystal materials.

[0009] This invention provides a high-throughput method for obtaining light intensity spectra of crystal materials, comprising the following steps:

[0010] Step 1: Obtain the preset lighting conditions. The preset lighting conditions include the different relative positions or angles of the lighting source, the sample under test, and the imaging system. Send the preset lighting conditions to the lighting equipment. When the lighting equipment executes the preset lighting conditions, a low spatial resolution intensity image sequence is acquired through the preset small numerical aperture optical imaging system to obtain the low spatial resolution intensity image sequence.

[0011] Step 1 Detailed Implementation:

[0012] Preset lighting conditions: By adjusting the relative positions or angles of the light source, the sample being tested, and the imaging system, various lighting scenarios can be generated. For example:

[0013] Keep the sample and imaging system fixed, and only change the incident angle of the illumination source (e.g., by lighting up LEDs at different positions one by one using an LED array).

[0014] Keep the sample and light source fixed, and adjust the acquisition angle of the imaging system.

[0015] Keep the light source and imaging system fixed, and rotate or translate the sample to change its orientation.

[0016] Image Acquisition: Using a small numerical aperture (NA) optical imaging system, a sequence of low spatial resolution intensity images is acquired under each preset illumination condition. The small NA system can cover a wider aperture angle range, ensuring high spectral resolution.

[0017] Step 1: Purpose

[0018] Frequency domain information coverage: Spectral information of different regions in the frequency domain corresponding to different lighting conditions, providing a data foundation for subsequent super-resolution reconstruction.

[0019] Guaranteed spectral resolution: The small NA system captures light intensity changes in more directions through a large aperture angle, improving the sensitivity of the spectrum to grain orientation.

[0020] Step 1 Technical Effects:

[0021] This addresses the contradiction between high spatial resolution and high spectral resolution in existing technologies, providing raw data that balances a large field of view and high sensitivity for subsequent steps.

[0022] Step 2: Process the low spatial resolution intensity image sequence using a pre-defined super-resolution wavefront reconstruction model to obtain a high spatial resolution complex amplitude field:

[0023] Step 2 Detailed Implementation:

[0024] Forward modeling: Based on the theory of light wavefront propagation, a mathematical model of the modulation of the light field by the sample under test is established. The complex amplitude field contains amplitude / phase information in the spatial domain and its frequency domain representation, and the two are related by Fourier transform.

[0025] Iterative reconstruction algorithm:

[0026] Initialization: Use the mean or interpolation result of the low-resolution image as the initial complex amplitude field estimate.

[0027] Frequency domain stitching: Using the spectrum of low-resolution images under different lighting conditions, the missing regions of high-resolution spectrum are gradually filled in the frequency domain.

[0028] Adaptive optimization: dynamically adjusts the iteration step size to balance convergence speed and stability; optionally, it jointly optimizes the transfer function of the imaging system (such as aberration correction).

[0029] Step 2's purpose:

[0030] Breaking the diffraction limit: By complementary splicing of frequency domain information, a complex amplitude field that surpasses the physical resolution limit of optical systems can be reconstructed.

[0031] Phase information extraction: The complex amplitude field contains phase information that is missing in traditional intensity images, providing additional features for grain boundary identification.

[0032] Step 2 Technical Effects:

[0033] To solve the problem of "inability to adapt to small grains" and achieve subwavelength level grain observation;

[0034] This provides high spatial resolution amplitude and phase data for subsequent steps, improving segmentation accuracy.

[0035] Step 3: Based on the high spatial resolution complex amplitude field, perform grain boundary identification on the high spatial resolution complex amplitude field to obtain a grain boundary map. Perform closure processing on the grain boundaries to segment multiple continuous regions, and regard the segmented different regions as different grains.

[0036] Step 3: Detailed Implementation Method

[0037] Grain boundary detection:

[0038] Amplitude and phase images are extracted from the complex amplitude field, respectively.

[0039] Perform Canny edge detection on both images to generate binarized edge maps.

[0040] Merge the two edge maps (logical OR operation) to mark all potential grain boundary locations.

[0041] Region growth segmentation:

[0042] Select a seed point in the grain boundary map (e.g., inside a known grain) and expand the region based on the continuity of non-grain boundary pixels.

[0043] Repeat the growth process until all grain regions are closed, generating an independent grain segmentation pattern.

[0044] Step 3:

[0045] Multi-feature fusion: The combined use of amplitude and phase information enhances the contrast between grain boundary and non-grain boundary regions.

[0046] High-resolution support: The high spatial resolution of the complex amplitude field reduces edge blurring and avoids erroneous merging of adjacent grains.

[0047] Step 3 Technical Effects:

[0048] This solves the problem of "limited grain segmentation accuracy" and significantly reduces the rate of missed detections / false detections.

[0049] It supports precise segmentation of tiny grains, such as those at the submicron level.

[0050] Step 4: Statistically analyze the light intensity of pixels contained in the grain under different lighting conditions. For each grain, iterate through the interpolated image sequence under each lighting condition. For each lighting condition, extract the pixel values ​​in the corresponding grain region and calculate the average intensity to form a light intensity spectrum describing the relationship between lighting conditions and grain light intensity.

[0051] Step 4: Detailed Implementation Steps

[0052] Calibration and calibration: Images are acquired using standard samples (such as uniform white boards) under different lighting conditions to eliminate the influence of lighting non-uniformity on light intensity measurement.

[0053] Image interpolation: Interpolates low-resolution intensity images to a high-resolution grid, aligned with the complex amplitude field space.

[0054] Light intensity statistics:

[0055] Pixel values ​​for each grain region are extracted based on the segmentation results.

[0056] The average light intensity of each grain under different lighting conditions is calculated and arranged according to the illumination angle and imaging angle to form a multidimensional light intensity spectrum.

[0057] Step 4's purpose:

[0058] Data consistency: Through calibration and interpolation, ensure spatial matching between light intensity data and high-resolution segmentation results.

[0059] High-throughput characterization: Multidimensional maps comprehensively reflect the light intensity response characteristics of grains under different illumination-observation combinations.

[0060] Step 4 Technical Effects:

[0061] To achieve "high throughput" output while meeting the requirements of high spatial resolution, high spectral resolution and large field of view;

[0062] It provides a high-precision data foundation for applications such as crystal orientation analysis and defect detection.

[0063] This invention overcomes the contradiction between spatial and spectral resolution in traditional methods by combining a small-NA imaging system with super-resolution wavefront reconstruction, and significantly improves grain segmentation accuracy by utilizing the phase information of the complex amplitude field. Steps 1-4 form a closed-loop process, requiring no additional hardware support from data acquisition to spectrum generation. High-throughput characterization is achieved solely through algorithmic innovation, providing an efficient and low-cost technical solution for crystal material research.

[0064] Furthermore, in the high-throughput method for obtaining light intensity spectra of crystal materials according to the present invention, step S1 includes:

[0065] The different relative positions or angular relationships among the illumination source, the sample being tested, and the imaging system include:

[0066] Keeping the sample and imaging system in the same position and orientation, adjust the position and angle of the illumination light according to the preset illumination light;

[0067] Keeping the position and orientation of the sample and the illumination light unchanged, adjust the position and angle of the imaging system according to the preset imaging system.

[0068] Keeping the imaging system and illumination light pose unchanged, the position and angle of the sample are changed by adjusting the angle of the sample according to the preset sample.

[0069] The preset lighting conditions are sent to the lighting equipment in an appropriate format, and the lighting equipment performs the corresponding lighting operation after receiving the preset conditions.

[0070] Furthermore, step S1 of the high-throughput crystal material intensity spectrum acquisition method of the present invention further includes...

[0071] After receiving preset lighting conditions, the lighting equipment performs corresponding lighting operations, including:

[0072] The lighting equipment adjusts the angle of the lighting light according to the preset lighting light;

[0073] While the lighting equipment is performing preset lighting conditions, images are acquired through a preset small numerical aperture optical imaging system; the above acquisition process is repeated under different preset lighting conditions to obtain intensity images with low spatial resolution.

[0074] Further, step S2 of the high-throughput crystal material intensity spectrum acquisition method of the present invention includes:

[0075] The forward model is used to analyze the modulation effect of the test sample on the light wavefront. The forward model characterizes the modulation characteristics of the sample through the complex amplitude field, which is correlated in the spatial domain and frequency domain through spectral transformation.

[0076] Based on the forward model, a super-resolution wavefront reconstruction algorithm is applied to reconstruct a complex amplitude field with high spatial resolution through initialization, iterative updates, and convergence judgment.

[0077] Furthermore, in the high-throughput method for obtaining the light intensity spectrum of crystal materials, step S2 includes:

[0078] The forward model is used to analyze the modulation effect of the test sample on the light wavefront. The forward model characterizes the modulation characteristics of the sample through the complex amplitude field, which is correlated in the spatial domain and frequency domain through spectral transformation.

[0079] Based on the forward model, a super-resolution wavefront reconstruction algorithm is applied to reconstruct a complex amplitude field with high spatial resolution through initialization, iterative updates, and convergence judgment.

[0080] Furthermore, the high-throughput method for obtaining light intensity spectra of crystal materials is characterized in that the forward model specifically includes:

[0081] The modulation effect of the test sample on the light wavefront is achieved through the complex amplitude field. Description, in which and For spatial coordinates, its spectral transformation is: ,in and Frequency domain coordinates;

[0082] No. The wave vector of the illumination light during secondary imaging is The complex amplitude field emitted from the sample surface is ,in, It is a spatial position vector. ,in They respectively represent the illumination wave vector at... direction and Components in direction.

[0083] Furthermore, the high-throughput method for obtaining the light intensity spectrum of crystal materials is characterized by further comprising:

[0084] Super-resolution wavefront reconstruction algorithm:

[0085] Reconstruct a high-resolution complex amplitude field by solving the following optimization problem:

[0086] The specific calculation steps are as follows:

[0087] initialization: The above initialization process is quite arbitrary; for example, The mean of all low-resolution images can be obtained or set through interpolation. Set to a constant of 0.

[0088] Furthermore, the high-throughput method for obtaining the light intensity spectrum of crystal materials is characterized by further comprising:

[0089] Iterative updates:

[0090] Generate the corresponding number Low-resolution complex amplitude field under different lighting conditions:

[0091]

[0092] Replace the amplitude component in the low-resolution complex amplitude field to match the intensity in the low-resolution image sequence described in step 1:

[0093]

[0094] Update the corresponding region of the high-resolution complex amplitude field spectrum:

[0095]

[0096] Repeat the above steps until all After frame processing is complete, a new iteration result is obtained: and .

[0097] Furthermore, in the high-throughput crystal material intensity spectrum acquisition method, the super-resolution wavefront reconstruction algorithm further includes:

[0098] During the iteration process, the reconstruction error or convergence rate is monitored in real time, and the step size α is dynamically adjusted. When the error cannot be effectively reduced, the step size is reduced, and when the convergence is good, the step size is increased.

[0099] Furthermore, the high-throughput crystal material light intensity spectrum acquisition method employs a high-order optimization method in the super-resolution wavefront reconstruction algorithm: during iterative updates, the objective function is approximately expanded to a second order, and the sample spectrum is updated using a global approximate Newton iteration, thereby reducing the number of iterations.

[0100] Furthermore, in the high-throughput method for obtaining the light intensity spectrum of crystal materials, step S3 includes:

[0101] The amplitude and phase components are extracted from the complex amplitude field with high spatial resolution, and the amplitude image and phase image are generated respectively.

[0102] The Canny edge detection algorithm was applied to the amplitude image and the phase image respectively to identify edge features in the amplitude image and the phase image;

[0103] Receive and set a threshold, perform binarization on the detected edges, and generate an edge image;

[0104] The binary edge images of the amplitude and phase images are processed so that if a pixel is marked as an edge in either image, it is also marked as an edge in the final grain boundary map, resulting in a grain boundary map containing grain boundary information. The boundary between different regions in the grain boundary map is the grain boundary.

[0105] Furthermore, in the high-throughput method for obtaining the light intensity spectrum of crystal materials, step S3 further includes:

[0106] In the grain boundary diagram, an initial seed point is selected, which is located inside or at the edge of a known grain;

[0107] Define growth criteria and perform region growth based on whether a pixel belongs to a grain boundary;

[0108] Starting from the seed point, the grain region is gradually expanded according to the growth criteria until all pixels that meet the conditions are included.

[0109] Repeat the above region growth process until all possible grains are segmented, generating a grain map containing different grain regions. In the grain map, the segmented different regions are regarded as different grains, resulting in a grain map containing different grain regions, with each grain region corresponding to an independent grain.

[0110] Furthermore, in the high-throughput method for obtaining the light intensity spectrum of crystal materials, step S4 includes:

[0111] Calibration was performed using standard samples under all preset lighting conditions;

[0112] Images of standard samples under different lighting conditions are acquired to characterize the brightness non-uniformity caused by the lighting conditions themselves.

[0113] Based on the calibration results, the low spatial resolution intensity image sequence is corrected to obtain the corrected low spatial resolution intensity image sequence.

[0114] Furthermore, in the high-throughput method for obtaining the light intensity spectrum of crystal materials, step S4 further includes:

[0115] The corrected low spatial resolution intensity image sequence is interpolated to obtain the interpolated image sequence.

[0116] For each grain, the interpolated images under all lighting conditions are traversed, and the pixel values ​​of the grain region are extracted when the quality of the interpolated image reaches the preset image standard.

[0117] The boundary of each grain is determined based on the grain diagram. In the interpolated image under each illumination condition, the pixel values ​​of the corresponding grain region are extracted to obtain the set of pixel values ​​of each grain under different illumination conditions.

[0118] Furthermore, in the high-throughput method for obtaining the light intensity spectrum of crystal materials, step S4 further includes:

[0119] The average intensity of the pixel value set of each grain under different lighting conditions is calculated. The average intensity includes the light intensity information of the grain under the preset lighting angle, and the average intensity value of each grain under different lighting conditions is obtained.

[0120] The average intensity values ​​of each grain under various illumination conditions are arranged according to the illumination angle and imaging angle to obtain a light intensity spectrum that includes the light intensity variation characteristics of all grains.

[0121] The advantages and beneficial effects of this invention are:

[0122] Compared with existing technologies, the technical solution of this invention has a higher information throughput and resolves the contradiction between high spatial resolution and high spectral resolution.

[0123] Specifically, this method is the first to apply super-resolution wavefront reconstruction technology to a light intensity spectrum acquisition system: using a small-NA imaging system, it acquires image sequences with high spectral resolution and a large field of view, but low spatial resolution. Then, super-resolution wavefront reconstruction technology stitches together the information in the frequency domain of the image sequences acquired under different illumination conditions to achieve high spatial resolution. While maintaining the advantages of high spectral resolution and a large field of view of the small-NA imaging system, it achieves the high spatial resolution effect of the large-NA imaging system, realizing for the first time the acquisition of light intensity spectra of high-throughput crystal materials.

[0124] This invention enables the acquisition of light intensity spectra of smaller grains. The application of super-resolution wavefront reconstruction technology allows the imaging results to break through the physical diffraction limit of the optical system and achieve higher optical resolution, thus enabling smaller grains to be effectively observed and identified.

[0125] This invention offers higher grain segmentation accuracy by employing super-resolution wavefront reconstruction technology to obtain a complex amplitude field with high spatial resolution. Compared to existing intensity spectrum acquisition methods that only provide low spatial resolution intensity information for grain boundary identification and grain segmentation, this method offers higher spatial resolution and can provide additional phase information, thus improving the accuracy of grain boundary identification and grain segmentation.

[0126] This invention requires no additional hardware configuration. The data required for both intensity spectrum acquisition and super-resolution wavefront reconstruction are intensity image sequences under different illumination conditions, thus exhibiting excellent consistency in system hardware configuration. The method fully leverages this consistency in system configuration, requiring no additional system hardware configuration for implementation. Attached Figure Description

[0127] Figure 1 This is a schematic diagram of a high-throughput method for obtaining light intensity spectra of crystal materials.

[0128] Figure 2 This is a schematic diagram of an existing technology for holding samples and imaging systems. Detailed Implementation

[0129] An embodiment of the present invention will be further described below with reference to the accompanying drawings.

[0130] Step 1: Acquire low spatial resolution intensity image sequences using a small numerical aperture (NA) optical imaging system under different lighting conditions. ,in This represents the total number of images. The different lighting conditions refer to the different relative positions or angular relationships between the lighting source, the sample under test, and the imaging system, and can be achieved in any of the following ways:

[0131] Keep the pose of the sample and imaging system unchanged, but change the position and angle of the illumination light;

[0132] Keeping the pose of the sample and the illumination light constant, change the position and angle of the imaging system;

[0133] Keep the orientation of the imaging system and illumination light constant, but change the position and angle of the sample being tested.

[0134] Appendix Figure 2 A prior art configuration is presented: keeping the sample and imaging system pose unchanged, changing the angle of the illumination light by lighting one LED at a different position in the LED array each time, and acquiring images through an optical imaging system to obtain the low-resolution intensity image sequence.

[0135] The purpose of step 1 is twofold: Firstly, images under different illumination conditions contain information about the sample at different locations in the frequency domain, providing information for super-resolution wavefront reconstruction in step 2.

[0136] On the other hand, images under different lighting conditions reflect the modulation effect of the sample on light under different lighting conditions and different reflection directions, providing information for the construction of the light intensity spectrum in step 4.

[0137] Step 2: Reconstruct the low spatial resolution intensity image sequence described in Step 1 using super-resolution wavefront reconstruction technology. After processing, a complex amplitude field with high spatial resolution is obtained. Specifically, it includes:

[0138] Forward model:

[0139] The modulation effect of the test sample on the light wavefront is achieved through the complex amplitude field. Description, in which and For spatial coordinates, its spectral transformation is: ,in and These are frequency domain coordinates.

[0140] No. The wave vector of the illumination light during secondary imaging is The complex amplitude field emitted from the sample surface is ,in, It is a spatial position vector. ,in They respectively represent the illumination wave vector at... direction and Components in direction.

[0141] The spectral transformation of the emitted complex amplitude field is as follows:

[0142] in, This represents spectral transformation.

[0143] The intensity image recorded by the imaging system is as follows:

[0144]

[0145] in Indicates the first Optical transfer function during sub-imaging.

[0146] The formula for the optical transfer function is as follows: Where circ is a circular domain function, and NA represents the numerical aperture of the imaging system. This represents system aberrations.

[0147] Super-resolution wavefront reconstruction algorithm:

[0148] Reconstruct a high-resolution complex amplitude field by solving the following optimization problem:

[0149] The specific calculation steps are as follows:

[0150] initialization: The above initialization process is quite arbitrary; for example, The mean of all low-resolution images can be obtained or set through interpolation. Set to a constant of 0.

[0151] Iterative updates:

[0152] Generate the corresponding number Low-resolution complex amplitude field under different lighting conditions:

[0153]

[0154] Replace the amplitude component in the low-resolution complex amplitude field to match the intensity in the low-resolution image sequence described in step 1:

[0155]

[0156] Update the corresponding region of the high-resolution complex amplitude field spectrum:

[0157]

[0158] Repeat the above steps until all After frame processing is complete, a new iteration result is obtained: and .

[0159] Convergence criterion: Check if the iteration results meet the convergence condition. If not, continue iterating until the iteration results meet the convergence condition, ultimately obtaining a complex amplitude field with high spatial resolution. An example of the above convergence condition is the number of iteration rounds. Reaching the predetermined maximum number of iterations .

[0160] The super-resolution wavefront reconstruction technique described in step 2 above can be combined with various improvement strategies to further enhance the robustness, efficiency, or accuracy of the reconstruction under conditions of noise, aberrations, or limited data acquisition. These improvements can be implemented independently or in combination, and include, but are not limited to, the following approaches (the following descriptions are for illustrative purposes only; in actual implementation, they can be flexibly selected or combined according to requirements):

[0161] Other formulas are used when updating the spectrum of high-resolution complex amplitude fields, for example:

[0162]

[0163] in, It is a custom update step size.

[0164] During the reconstruction process, the transfer function of the imaging system is simultaneously estimated. In the iterative process of the reconstruction algorithm described above, a joint solution mechanism for the optical transfer function of the imaging system is introduced. Specifically:

[0165] The optical transfer function of the imaging system Treat it as an unknown quantity and the sample spectrum in the objective function. Joint optimization should be carried out together;

[0166] In each iteration, the amplitude and / or phase distribution of the transfer function is updated based on the intensity observations and the currently estimated sample spectrum;

[0167] By continuously adjusting the transfer function, a more accurate estimate of aberrations or other mismatches in a real optical system can be obtained, thereby reducing reliance on prior information about the system.

[0168] Adaptively adjust the step size to enhance robustness to noise;

[0169] In the presence of noise, iterative step size must be carefully selected; otherwise, oscillations or failure to converge may occur. Therefore, the step size can be dynamically adjusted with each iteration update:

[0170] make For the first The step size of the next iteration;

[0171] During the iteration process, the reconstruction error or convergence rate is monitored in real time. When it is found that the error cannot be effectively reduced, the current step size is reduced. When convergence is good, the scale can be appropriately increased. This balances convergence speed and stability.

[0172] Accelerate convergence using higher-order or approximate second-order methods;

[0173] In the iterative super-resolution wavefront reconstruction process described above, higher-order optimization strategies can be employed to update the sample spectrum. For example, the objective function can be approximated with a second-order expansion during iterative updates, or a global approximation of Newton's iterations can be used to solve the problem, thereby approximating the optimal solution more quickly while maintaining high computational efficiency. The general idea can be summarized as follows:

[0174] In each iteration, a local second-order approximation is made to the objective function, introducing a matrix similar to the Hessian matrix or its approximate form.

[0175] By appropriately sampling and regularizing in the frequency or spatial domains, the excessively large Hessian matrix can be avoided through explicit computation.

[0176] Achieve The search direction is updated more accurately, thereby reducing the number of iterations or enhancing robustness under high noise conditions.

[0177] The purpose of step 2:

[0178] First, super-resolution wavefront reconstruction technology can overcome the constraint between spatial resolution and spectral resolution. High spectral resolution images can be acquired using a small-NA imaging system, and high spatial resolution can be achieved through super-resolution wavefront reconstruction. Ultimately, this results in higher information throughput.

[0179] Second, super-resolution wavefront reconstruction technology can improve the optical resolution of images, thereby enabling the observation of smaller grains and breaking through the limitations on grain size in existing light intensity spectrum acquisition methods.

[0180] Third, the complex amplitude field can be obtained through super-resolution wavefront reconstruction technology. Compared with existing intensity spectrum acquisition methods that can only provide intensity information for grain boundary identification and grain segmentation, this method can provide additional phase information, thereby improving the accuracy of grain boundary identification and grain segmentation in step 3.

[0181] Step 3: Based on the high spatial resolution complex amplitude field obtained in Step 2 By analyzing the phase and amplitude information in the complex amplitude field, the location of grain boundaries is identified, and different grains are segmented based on the identified grain boundaries.

[0182] The specific processing method for step 3:

[0183] Grain boundary identification: Take the amplitude portion of the complex amplitude field obtained in step 2 respectively. and phase part The Canny edge detection algorithm is then applied to obtain the edge image. By setting appropriate thresholds, binarizing the data, and then performing an OR operation on the two values, a grain boundary map can be obtained. ,in This indicates a binarization operation. Representation or operation.

[0184] Grain segmentation: Processing grain boundary maps using a region-growing-based segmentation algorithm. Different regions are treated as different grains, and the segmentation of these different grains is achieved to obtain a grain map. .

[0185] Step 4: Based on the low-resolution intensity image sequence described in Step 1 and the segmented grains from Step 3, construct the intensity spectrum of different grains. For example, for each grain, calculate the average intensity of the pixels contained in the grain under different illumination conditions to form an intensity spectrum.

[0186] The specific processing method for step 4:

[0187] Calibration and calibration: Under the different lighting conditions described in step 1, the system is calibrated using standard samples (such as a uniformly colored white board) to obtain standard sample images under different lighting conditions, which are used to characterize the brightness non-uniformity caused by the lighting conditions themselves. Based on the calibration results, the low-resolution intensity image sequence described in step 1 is corrected to avoid the influence of brightness differences caused by lighting non-uniformity.

[0188] Image interpolation: The low-resolution intensity image sequence described in step 1 is interpolated to match the pixel size of the high-resolution complex amplitude field described in step 2, resulting in an interpolated image sequence. .

[0189] Light intensity map construction: For each grain, the interpolated image sequence under each illumination condition is traversed, the pixel values ​​within the corresponding grain region are extracted, and the average intensity of the pixel values ​​is calculated. The average intensity values ​​of each grain under each illumination condition are arranged according to the brightness angle and imaging angle to form a light intensity map. Each light intensity map represents the light intensity variation characteristics of a grain.

[0190] Please see the appendix Figure 1 The present invention provides a method for obtaining light intensity spectra of high-throughput crystal materials, comprising the following steps:

[0191] Step 1: Acquire a sequence of low spatial resolution intensity images using a small numerical aperture (NA) optical imaging system under different lighting conditions;

[0192] Step 2: Process the low spatial resolution intensity image sequence using super-resolution wavefront reconstruction technology to obtain a high spatial resolution complex amplitude field;

[0193] Step 3: Perform grain boundary identification and grain segmentation based on the high spatial resolution complex amplitude field;

[0194] Step 4: Calculate the light intensity of the pixels contained in the grain under different lighting conditions and construct a light intensity map.

[0195] The core advantage of this method is its "high throughput," which encompasses three aspects: high spatial resolution, high spectral resolution, and a large field of view. Steps 1 and 2 are crucial for achieving these advantages.

[0196] This invention provides a high-throughput method for acquiring light intensity spectra of crystal materials, aiming to address several technical pain points in the prior art. The following are the detailed steps and logical sequence by which this invention solves these technical pain points:

[0197] Spatial resolution and spectral resolution are mutually limiting: In existing methods, the high cutoff frequency of large numerical aperture (NA) imaging systems means high spatial resolution, but the imaging light intensity does not change significantly when illumination conditions change, resulting in low spectral resolution.

[0198] Unable to adapt to small grains: Existing methods can only use small NA objectives for imaging, which limits spatial resolution and makes it impossible to detect small grains.

[0199] Limited grain segmentation accuracy: The low image resolution and the fact that it is only an intensity image without phase information result in limited information for grain boundary identification and grain segmentation.

[0200] The solution of this invention:

[0201] Low spatial resolution intensity image sequences were acquired using a small numerical aperture (NA) optical imaging system under different illumination conditions. With proper illumination settings (including the relative positions or angles of the illumination source, the sample under test, and the imaging system), low spatial resolution intensity image sequences were successfully acquired.

[0202] A pre-defined super-resolution wavefront reconstruction model was used to process low spatial resolution intensity image sequences to obtain high spatial resolution complex amplitude fields. With reasonable adjustments to model parameters (such as iteration step size and convergence conditions), the high spatial resolution complex amplitude field was successfully reconstructed.

[0203] Grain boundary identification and grain segmentation are performed based on a complex amplitude field with high spatial resolution to obtain a grain boundary map. Grain boundaries are then closed to segment multiple continuous regions, and these different regions are considered as different grains. With proper application of grain boundary identification algorithms (such as Canny edge detection) and segmentation algorithms (such as region growing), grain boundaries are successfully identified and different grains are segmented.

[0204] The light intensity of pixels within a grain under different lighting conditions was statistically analyzed to form a light intensity spectrum. With proper calibration, correction, and interpolation processing of the low-resolution intensity image sequence, the average light intensity of each grain under different lighting conditions was successfully calculated, forming a light intensity spectrum.

[0205] This invention resolves the trade-off between high spatial resolution and high spectral resolution, enabling high-throughput acquisition of light intensity spectra of crystal materials. The application of super-resolution wavefront reconstruction technology allows imaging results to break through the physical diffraction limit of optical systems, achieving higher optical resolution and thus enabling the observation of smaller grains.

[0206] Providing a complex amplitude field with high spatial resolution, including additional phase information, helps improve the accuracy of grain boundary identification and grain segmentation. Both the intensity spectrum acquisition and super-resolution wavefront reconstruction techniques require intensity image sequences under different illumination conditions, exhibiting excellent consistency in system hardware configuration.

[0207] Through the above steps, the present invention effectively solves the technical pain points existing in the background art and provides a more accurate and efficient method for obtaining light intensity spectrum for the research of crystal materials.

Claims

1. A method for obtaining high-throughput light intensity spectra of crystal materials, comprising the following steps: Step 1: Obtain the preset lighting conditions. The preset lighting conditions include the different relative positions or angles of the lighting source, the sample under test, and the imaging system. Send the preset lighting conditions to the lighting equipment. When the lighting equipment executes the preset lighting conditions, a low spatial resolution intensity image sequence is acquired through the preset small numerical aperture optical imaging system to obtain the low spatial resolution intensity image sequence. Step 2: Process the low spatial resolution intensity image sequence using a preset super-resolution wavefront reconstruction model to obtain a high spatial resolution complex amplitude field; Step 3: Based on the high spatial resolution complex amplitude field, perform grain boundary identification on the high spatial resolution complex amplitude field to obtain a grain boundary map. Perform closure processing on the grain boundaries to segment multiple continuous regions, and regard the segmented different regions as different grains. Step 4: Statistically analyze the light intensity of pixels contained in the grain under different lighting conditions. For each grain, iterate through the interpolated image sequence under each lighting condition. For each lighting condition, extract the pixel values ​​in the corresponding grain region and calculate the average intensity to form a light intensity spectrum describing the relationship between lighting conditions and grain light intensity.

2. The method for obtaining high-throughput light intensity spectra of crystal materials as described in claim 1, characterized in that, Step 1 includes: The different relative positions or angular relationships among the illumination source, the sample being tested, and the imaging system include: Keeping the sample and imaging system in the same position and orientation, adjust the position and angle of the illumination light according to the preset illumination light; Keeping the position and orientation of the sample and the illumination light unchanged, adjust the position and angle of the imaging system according to the preset imaging system. Keeping the imaging system and illumination light pose unchanged, the position and angle of the sample are changed by adjusting the angle of the sample according to the preset sample. The preset lighting conditions are sent to the lighting equipment in an appropriate format, and the lighting equipment performs the corresponding lighting operation after receiving the preset conditions.

3. The method for obtaining high-throughput light intensity spectra of crystal materials as described in claim 2, characterized in that, Step 1 also includes After receiving preset lighting conditions, the lighting equipment performs corresponding lighting operations, including: The lighting equipment adjusts the angle of the lighting light according to the preset lighting light; While the lighting equipment is performing preset lighting conditions, images are acquired through a preset small numerical aperture optical imaging system; the above acquisition process is repeated under different preset lighting conditions to obtain intensity images with low spatial resolution.

4. The method for obtaining high-throughput light intensity spectra of crystal materials as described in claim 1, characterized in that, Step 2 includes: The forward model is used to analyze the modulation effect of the test sample on the light wavefront. The forward model characterizes the modulation characteristics of the sample through the complex amplitude field, which is correlated in the spatial domain and frequency domain through spectral transformation. Based on the forward model, a super-resolution wavefront reconstruction algorithm is applied to reconstruct a complex amplitude field with high spatial resolution through initialization, iterative updates, and convergence judgment.

5. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 4, characterized in that, The forward model specifically includes: The modulation effect of the test sample on the light wavefront is achieved through the complex amplitude field. Description, in which and For spatial coordinates, its spectral transformation is: ,in and Frequency domain coordinates; No. The wave vector of the illumination light during secondary imaging is The complex amplitude field emitted from the sample surface is , ,in, It is a spatial position vector. ,in They respectively represent the illumination wave vector at... direction and Components in direction.

6. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 4, characterized in that, The super-resolution wavefront reconstruction algorithm reconstructs a high-resolution complex amplitude field by solving the following optimization problem: 。 7. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 6, characterized in that, The specific calculation steps of the super-resolution wavefront reconstruction algorithm are as follows: initialization: ,in The mean of all low-resolution images can be obtained or set through interpolation. Set to a constant of 0: Iterative updates: Generate the corresponding number Low-resolution complex amplitude field under different lighting conditions: ； Replace the amplitude component in the low-resolution complex amplitude field to match the low spatial resolution intensity image sequence described in step 1: ； Update the corresponding region of the high-resolution complex amplitude field spectrum: ； Repeat the above steps until all After frame processing is complete, a new iteration result is obtained: and .

8. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 6, characterized in that, The super-resolution wavefront reconstruction algorithm also includes: During the iteration process, the reconstruction error or convergence rate is monitored in real time, and the step size α is dynamically adjusted. When the error cannot be effectively reduced, the step size is reduced, and when the convergence is good, the step size is increased.

9. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 6, characterized in that, The super-resolution wavefront reconstruction algorithm employs a high-order optimization method: During iterative updates, the objective function is expanded to an approximate second order, and the sample spectrum is updated using a global approximate Newton iteration, reducing the number of iterations.

10. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 1, characterized in that, Step 3 includes: The amplitude and phase components are extracted from the complex amplitude field with high spatial resolution, and the amplitude image and phase image are generated respectively. The Canny edge detection algorithm was applied to the amplitude image and the phase image respectively to identify edge features in the amplitude image and the phase image; Receive and set a threshold, perform binarization on the detected edges, and generate an edge image; The binary edge images of the amplitude and phase images are processed so that if a pixel is marked as an edge in either image, it is also marked as an edge in the final grain boundary map, resulting in a grain boundary map containing grain boundary information. The boundary between different regions in the grain boundary map is the grain boundary.

11. The method for obtaining high-throughput light intensity spectra of crystal materials as described in claim 10, characterized in that, Step 3 further includes: In the grain boundary diagram, an initial seed point is selected, which is located inside or at the edge of a known grain; Define growth criteria and perform region growth based on whether a pixel belongs to a grain boundary; Starting from the seed point, the grain region is gradually expanded according to the growth criteria until all pixels that meet the conditions are included. Repeat the above region growth process until all possible grains are segmented, generating a grain map containing different grain regions. In the grain map, the segmented different regions are regarded as different grains, resulting in a grain map containing different grain regions, with each grain region corresponding to an independent grain.

12. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 1, characterized in that, Step 4 includes: Calibration was performed using standard samples under all preset lighting conditions; Images of standard samples under different lighting conditions are acquired to characterize the brightness non-uniformity caused by the lighting conditions themselves. Based on the calibration results, the low spatial resolution intensity image sequence is corrected to obtain the corrected low spatial resolution intensity image sequence.

13. The method for obtaining high-throughput light intensity spectra of crystal materials as described in claim 12, characterized in that, Step 4 also includes: The corrected low spatial resolution intensity image sequence is interpolated to obtain the interpolated image sequence. For each grain, the interpolated images under all lighting conditions are traversed, and the pixel values ​​of the grain region are extracted when the quality of the interpolated image reaches the preset image standard. The boundary of each grain is determined based on the grain diagram. In the interpolated image under each illumination condition, the pixel values ​​of the corresponding grain region are extracted to obtain the set of pixel values ​​of each grain under different illumination conditions.

14. The method for obtaining high-throughput light intensity spectra of crystal materials according to claim 13, characterized in that, Step 4 also includes: The average intensity of the pixel value set of each grain under different lighting conditions is calculated. The average intensity includes the light intensity information of the grain under the preset lighting angle, and the average intensity value of each grain under different lighting conditions is obtained. The average intensity values ​​of each grain under various illumination conditions are arranged according to the illumination angle and imaging angle to obtain a light intensity spectrum that includes the light intensity variation characteristics of all grains.

Images