Parallel scanning imaging method based on planar super-resolution lens array

Abstract

The application relates to a parallel scanning imaging method based on a planar super-resolution lens array, and relates to the technical field of fluorescence microscopic imaging, wherein the method comprises the following steps: placing a biological sample to be measured on a sample stage; controlling the sample stage to move the biological sample to be measured, performing full-field pre-scanning on the biological sample to be measured, synchronously collecting images of the biological sample to be measured during the full-field pre-scanning by means of an image collection module, and obtaining a pre-scanning panoramic image; dividing the pre-scanning panoramic image into a plurality of blocks, dividing different blocks into fine scanning areas and coarse scanning areas, and finally obtaining a super-resolution image corresponding to the fine scanning areas and a low-resolution image corresponding to the coarse scanning areas; and fusing the super-resolution image and the low-resolution image to obtain a fluorescence image of the biological sample to be measured. The application realizes the imaging requirements of the biological sample in super-resolution observation and low-exposure protection.

Description

Technical Field

[0001] This invention relates to the field of fluorescence microscopy imaging technology, and in particular to a parallel scanning imaging method based on a planar super-resolution lens array. Background Technology

[0002] With the development of micro-nano technology, planar optical lenses have become a research hotspot. While metasurface lenses offer flexible design, they are limited by the wavefront reconstruction principle and cannot overcome the diffraction limit; furthermore, the precision of large-area fabrication is difficult to control. Planar lenses based on the super-oscillatory principle can achieve far-field super-resolution by modulating the light field through microstructures, providing core support for imaging systems. How to leverage the advantages of super-oscillatory lenses, combined with system and algorithm optimization, to balance the contradiction between super-resolution, low exposure, and high scanning efficiency is currently a key bottleneck in the field.

[0003] This contradiction is particularly prominent in the field of biological fluorescence microscopy. Life science research increasingly seeks in-situ, real-time observation of living samples and dynamic physiological processes, which places extremely high demands on imaging technology: on the one hand, it requires super-resolution capabilities to resolve subcellular structures, the localization and interactions of biomolecules; on the other hand, it requires extremely low light doses to avoid photobleaching of fluorescent labels and fatal damage to living samples due to light source toxicity, thus maintaining their physiological activity. Current mainstream super-resolution fluorescence imaging techniques, including stimulated emission depletion microscopy (STED), stochastic optical reconstruction microscopy (STORM), and structured illumination microscopy (SIM), mainly rely on the effect of fluorescent molecular clusters. Although they can achieve imaging beyond the diffraction limit in the far field, they often require high-power excitation light for long-term scanning, making it difficult to balance speed and activity maintenance. While ordinary wide-field illumination microscopy has a high frame rate, it is limited by the diffraction limit and requires high illumination uniformity, making it difficult to achieve local super-resolution at low doses.

[0004] Furthermore, existing adaptive scanning imaging technologies employ low-resolution preview and local high-resolution scanning modes: target regions are identified through image entropy or deep learning, but relying on ordinary lenses for single-lens moving scans leads to an imbalance between resolution and efficiency. The main reason is that traditional refractive optical lenses lack super-resolution capabilities, and the introduction of metasurface lenses is still limited by the diffraction limit. In addition, metasurface lenses, due to their subwavelength linewidths, are difficult to fabricate at the wafer level and at low cost. Planar super-oscillating lenses, which can achieve focused imaging beyond the diffraction limit in the far field or even the ultra-far field without the aid of subwavelength linewidth structures, are gradually attracting researchers' attention. However, scanning imaging based on a single planar super-oscillating lens also suffers from low scanning efficiency and cannot perform efficient parallel scanning imaging of biological fluorescent samples.

[0005] In summary, existing super-resolution imaging techniques (such as single-mode scanning like STED, STORM, and SIM) require multiple scans of the sample and the use of biofluorescent molecules and complex image processing algorithms to achieve submicron resolution. This leads to a surge in exposure times for biological samples, causing phototoxicity and photobleaching effects. Meanwhile, low-exposure techniques rely on diffraction-limited traditional refractive optical lenses, which cannot achieve submicron-level super-resolution observation. In short, current methods cannot simultaneously achieve both super-resolution image acquisition and low-exposure damage when performing fluorescence microscopy. Summary of the Invention

[0006] Therefore, the technical problem to be solved by the present invention is to overcome the problem in the prior art that it is impossible to simultaneously take into account super-resolution image acquisition and low exposure damage when performing fluorescence microscopy.

[0007] To address the aforementioned technical problems, this invention provides a parallel scanning imaging method based on a planar super-resolution lens array, comprising:

[0008] Step S1: Place the biological sample to be tested on the stage;

[0009] Step S2: Control the stage to move the biological sample to be tested, perform a full-field pre-scan of the biological sample to be tested, and simultaneously acquire images through the image acquisition module while performing the full-field pre-scan of the biological sample to be tested to obtain a pre-scan panoramic image.

[0010] Step S3: Divide the pre-scanned panoramic image into several blocks, and divide the different blocks into fine scan areas and coarse scan areas, and finally obtain the super-resolution image corresponding to the fine scan area and the low-resolution image corresponding to the coarse scan area.

[0011] Step S4: Fuse the super-resolution image and the low-resolution image to obtain the fluorescence image of the biological sample to be tested.

[0012] In one embodiment of the present invention, step S3, which divides the pre-scanned panoramic image into several blocks and further divides the different blocks into fine-scan regions and coarse-scan regions, and finally obtains the super-resolution image corresponding to the fine-scan region and the low-resolution image corresponding to the coarse-scan region, includes the following method:

[0013] Step S301: Divide the pre-scanned panoramic image into n×m blocks, where n is the number of rows in the block and m is the number of columns in the block;

[0014] Step S302: Calculate the interest score for each block;

[0015] Step S303: Set a threshold for interest scores, mark blocks with interest scores higher than the threshold as fine scan areas, and mark blocks with interest scores lower than the threshold as coarse scan areas;

[0016] Step S304: The stage moves to perform super-resolution scanning on the fine scanning area of ​​the biological sample and low-exposure scanning on the coarse scanning area. During the scanning process, the image acquisition module acquires the super-resolution image corresponding to the super-resolution scan and the low-resolution image corresponding to the low-exposure scan.

[0017] In one embodiment of the present invention, step S302 calculates the interest score for each block, expressed as:

[0018] ;

[0019] ;

[0020] in, For the interest score of the block, Based on interest scores, For boundary interest scores, , , The first, second, and third weight coefficients are given, and they satisfy the following conditions: ;

[0021] The normalized image entropy is expressed as:

[0022] ;

[0023] ;

[0024] ;

[0025] in, The grayscale information entropy of an image block. grayscale value The probability of occurrence within a block. The maximum possible entropy for 8-bit grayscale;

[0026] To normalize the texture complexity, it is expressed as:

[0027] ;

[0028] ;

[0029] ;

[0030] in, For all preset angles of the image block Correspondence Texture complexity obtained by taking the average, for The maximum value in, Regarding the preset angle The complexity, Preset angle And a certain pixel pair under step size The probability of occurrence in all pixel pairs, preset angle For use in determining reference pixels adjacent pixels azimuth angle, Preset angle The number of;

[0031] The normalized target structure pixel ratio is expressed as:

[0032] = ;

[0033] ;

[0034] in, The number of target pixels after threshold segmentation. This represents the total number of pixels in the block.

[0035] In one embodiment of the present invention, the boundary interest score Represented as:

[0036] ;

[0037] in, The biological boundary enhancement coefficient;

[0038] The contribution value to the biological boundary is expressed as:

[0039] ;

[0040] in, The length of the continuous boundary extracted by Canny edge detection. The length of the block's diagonal. This is the boundary continuity factor.

[0041] In one embodiment of the present invention, step S303 further includes:

[0042] Based on the interest score above a threshold, the scan priority is assigned to the fine scan region; the higher the interest score, the higher the scan priority.

[0043] In one embodiment of the present invention, in step S304:

[0044] When the stage performs super-resolution scanning of the fine scanning area of ​​the biological sample according to the optimal scanning path, the stage moves by a step length, and the image acquisition module acquires the super-resolution image with a first exposure.

[0045] When the stage performs low-exposure scanning of the coarse scanning area of ​​the biological sample according to the optimal scanning path, the stage moves with a second step size, and the image acquisition module acquires a low-resolution image with a second exposure.

[0046] Wherein, the first step length is less than the second step length, and the first exposure amount is greater than the second exposure amount.

[0047] In one embodiment of the present invention, the method for fusing the super-resolution image and the low-resolution image in step S4 to obtain a fluorescence image of the biological sample to be tested includes:

[0048] Step S401: By matching block coordinates, the super-resolution image and the low-resolution image are stitched together to form a continuous image covering the entire field of view;

[0049] Step S402: Enhance the details of the super-resolution image region in the full-view continuous image and reduce the noise in the low-resolution image region to obtain the fluorescence image of the biological sample to be tested.

[0050] In one embodiment of the present invention, the method for enhancing the details of the super-resolution image region in the full-view continuous image in step S402 includes:

[0051] Based on basic interest score Boundary interest score Build Enhanced Weights , As the first coefficient, As the second coefficient, and Range 0~1; The Sobel algorithm is used to search for the structural edge pixels of biological cells in the super-resolution image, and the gray values ​​of the biological cell edge pixels are multiplied by (1+). )% to achieve enhanced details.

[0052] In one embodiment of the present invention, the method for enhancing the details of the super-resolution image region in the full-view continuous image in step S402 includes:

[0053] The fluorescence intensity of all biological cells in the super-resolution image is obtained. The fluorescence intensity of all biological cells is averaged and then multiplied by a preset compensation parameter to obtain a fluorescence enhancement compensation value. The fluorescence enhancement compensation value is superimposed on the area of ​​biological cells in the super-resolution image where the fluorescence intensity of biological cells is less than 200 ADU to achieve detail enhancement.

[0054] In one embodiment of the present invention, the method for noise reduction of the region where the low-resolution image is located in step S402 includes:

[0055] The Canny edge detection algorithm is used to identify the regions containing biological cells in the low-resolution image. For pixel blocks between adjacent biological cells, a circular region is constructed based on the minimum straight-line distance between the pixel blocks of adjacent biological cells. A noise reduction weight is then used to reduce the noise reduction intensity of this circular region. The noise reduction weight... Represented as:

[0056] ;

[0057] in, For noise reduction constant, The noise reduction coefficient is... To enhance weight and , As the first coefficient, The second coefficient;

[0058] The low-resolution image is denoised using Gaussian filtering. During the denoising process, the low-resolution image is divided into two parts: the first part is the low-resolution image excluding the annular region, which is then denoised using Gaussian filtering; the second part is the annular region, where the standard deviation of the Gaussian function is replaced with denoising weights. An improved Gaussian filter is obtained, and noise reduction in the annular region is achieved through the improved Gaussian filter.

[0059] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0060] The parallel scanning imaging system based on a planar super-resolution lens array constructed in this invention effectively breaks through the technical bottleneck of "the inability to balance super-resolution observation and low-exposure protection", and effectively avoids the inactivation of biological samples.

[0061] The parallel scanning imaging method based on a planar super-resolution lens array constructed in this invention divides the pre-scanned panoramic image into blocks. By constructing interest scores, different blocks are divided into fine-scanning regions and coarse-scanning regions. The interest scores take into account the actual scene and can achieve accurate division, ensuring the accuracy of the subsequently generated fluorescence images.

[0062] When fusing super-resolution and low-resolution images, this invention enhances the details of the super-resolution image region while reducing noise in the low-resolution image region, ensuring that the final fused image has more details and is more consistent with the actual scene. Attached Figure Description

[0063] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0064] Figure 1 This is a schematic diagram of the parallel scanning imaging system architecture based on a planar super-resolution lens array according to the present invention;

[0065] Figure 2 This is a schematic diagram of the objective lens switching module structure between the stage and the image acquisition module of the present invention;

[0066] Figure 3 This is a schematic diagram of the lens array of the present invention;

[0067] Figure 4 This is a flowchart of the parallel scanning imaging method based on a planar super-resolution lens array according to the present invention;

[0068] Figure 5 This is a flowchart of the method for partitioning, setting fine scanning areas, and coarse scanning areas based on a pre-scanned panoramic image according to the present invention.

[0069] Explanation of the reference numerals in the accompanying drawings: 1. Laser; 2. Beam expander; 3. Collimating lens; 4. Lens array; 5. Biological sample to be tested; 6. Stage; 7. Image acquisition module; 8. Objective lens switching module. Detailed Implementation

[0070] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0071] Example 1

[0072] Reference Figure 1 As shown, the present invention relates to a parallel scanning imaging system based on a planar super-resolution lens array, comprising an optical emission module, a lens array 4, a stage 6, and an image acquisition module 7 arranged sequentially.

[0073] The optical emission module includes a laser 1, a beam expander 2, and a collimating lens 3 arranged in sequence. The laser 1 is used to provide a laser source for fluorescence microscopy imaging. The beam expander 2 is used to expand the laser source to obtain expanded beam light. The collimating lens 3 is used to convert the expanded beam light into parallel light and then incident it onto the lens array 4.

[0074] The stage 6 is used to carry the biological sample 5 to be tested (such as live cells or tissue sections) and move the biological sample 5 to be tested to the lens array 4.

[0075] The light beam passing through the lens array 4 illuminates the biological sample 5 to be tested on the stage 6;

[0076] The image acquisition module 7 is used to capture images of the biological sample 5 after it has been irradiated by the lens array 4, and to obtain digital images of the biological sample 5.

[0077] Further, please refer to Figure 3 The lens array 4 includes a substrate 401 and a lens structure layer 402 disposed on the substrate 401; the lens structure layer 402 includes a plurality of super-oscillating micro-nano lenses 403, and the super-oscillating micro-nano lenses 403 adopt micro-nano optical structures.

[0078] In this embodiment, the lens array 4 integrates multiple super-oscillating micro / nano lenses 403. Please refer to the details below. Figure 3 (b) For details on the structure of each super-oscillating micro / nano lens 403, please refer to [link / reference]. Figure 3 (a), Figure 3 In (a), h1 is the thickness of substrate 401, and h2 is the thickness of lens structure layer 402. Array planar layout ( Figure 3 (b): Lens array 4 is distributed in a matrix in the XY plane (the figure shows a 2×2 array, which can be expanded to an N×M scale, where N is the number of lens rows and M is the number of lens columns).

[0079] Furthermore, in the lens array 4 distributed in an N×M matrix in the XY plane, the adjacent super-oscillating micro-nano lenses 403 are set with different spacings according to different biological samples 5 to be tested, for example, the spacing range is 10-30um.

[0080] Please see Figure 2 This embodiment also includes an objective lens switching module 8, which is disposed between the stage 6 and the image acquisition module 7. The objective lens switching module 8 is equipped with objective lenses of different magnifications. The objective lens switching module 8 achieves super-resolution scanning or low-exposure scanning of the biological sample 5 under test by switching between objective lenses of different magnifications. The objective lens switching module 8 adopts a rotary structure driven by a micro stepper motor, with a switching angle of 30°, a driving voltage of 12V, a switching time of ≤8ms, and a positioning accuracy of ±0.01mm (ensuring that the coaxiality error between the lens optical axis and the optical path is ≤0.02mm after switching). It achieves the switching of objective lenses of different magnifications by receiving TTL level signals.

[0081] Furthermore, the stage 6 has an XY motorized displacement adjustment function (response time ≤ 5ms), which can accurately move the biological sample 5 to be tested according to the scanning command.

[0082] Furthermore, laser 1 can be a narrow-linewidth laser with a fluorescence excitation wavelength.

[0083] Furthermore, the image acquisition module 7 is a camera, specifically a high-sensitivity fluorescence camera, such as an sCMOS camera.

[0084] Example 2

[0085] Please see Figure 4 This embodiment provides a parallel scanning imaging method based on a planar super-resolution lens array, employing the parallel scanning imaging system based on a planar super-resolution lens array as described in Embodiment 1, including:

[0086] Step S1: Place the biological sample 5 to be tested onto the stage 6;

[0087] Step S2: Control the stage 6 to move the biological sample 5 to be tested, perform a full-field pre-scan of the biological sample 5 to be tested, and simultaneously acquire images through the image acquisition module 7 while performing the full-field pre-scan of the biological sample 5 to obtain a pre-scan panoramic image.

[0088] Step S3: Divide the pre-scanned panoramic image into several blocks, and divide the different blocks into fine scan areas and coarse scan areas, and finally obtain the super-resolution image corresponding to the fine scan area and the low-resolution image corresponding to the coarse scan area.

[0089] Step S4: Fuse the super-resolution image and the low-resolution image to obtain the fluorescence image of the biological sample 5 to be tested.

[0090] Further, please refer to Figure 5 Step S3 divides the pre-scanned panoramic image into several blocks, and further divides the different blocks into fine-scan regions and coarse-scan regions, ultimately obtaining the super-resolution image corresponding to the fine-scan region and the low-resolution image corresponding to the coarse-scan region. The method includes:

[0091] Step S301: Divide the pre-scanned panoramic image into n×m blocks, where n is the number of rows in the block and m is the number of columns in the block;

[0092] Step S302: Calculate the interest score for each block;

[0093] Step S303: Set a threshold for the interest score, and mark the blocks with interest scores higher than the threshold as fine scan regions (requiring super-resolution observation, such as intracellular organelle distribution areas); mark the blocks with interest scores lower than the threshold as coarse scan regions (requiring only low resolution, such as extracellular culture medium areas).

[0094] Step S304: The stage 6 moves to perform super-resolution scanning on the fine scanning area of ​​the biological sample 5 and low-exposure scanning on the coarse scanning area. During the scanning process, the image acquisition module 7 acquires the super-resolution image corresponding to the super-resolution scan and the low-exposure image corresponding to the low-exposure scan.

[0095] Further, step S302 calculates the interest score for each block, expressed as:

[0096] ;

[0097] ;

[0098] in, For the interest score of the block, Based on interest scores, For boundary interest scores, , , The first, second, and third weight coefficients are given, and they satisfy the following conditions: .

[0099] The normalized image entropy is expressed as:

[0100] ;

[0101] ;

[0102] ;

[0103] in, The grayscale information entropy of an image block. grayscale value The probability of occurrence within a block. The maximum possible entropy for 8-bit grayscale.

[0104] To normalize the texture complexity, it is expressed as:

[0105] ;

[0106] ;

[0107] ;

[0108] in, For all preset angles of the image block Correspondence Texture complexity obtained by taking the average, for The maximum value in, Regarding the preset angle The complexity, Preset angle And a certain pixel pair under step size The probability of occurrence in all pixel pairs, preset angle For use in determining reference pixels adjacent pixels azimuth angle, Preset angle The number of elements. It should be noted that the texture complexity in this embodiment... Using contrast, pixel pairs This can be understood as: for example, adjacent pixels. For pixels To the right (i.e., angle) Take the neighboring pixels with a 0-degree angle and a step size of 1 pixel, and then... and Form a pixel pair In this embodiment, a preset angle is used. We select 0 degrees (horizontal to the right), 45 degrees (top right diagonal), 90 degrees (vertical upward), and 135 degrees (top left diagonal), and then obtain the image blocks at the corresponding angles. Four angles corresponding Taking the average yields the final result. .

[0109] The normalized target structure pixel ratio is expressed as:

[0110] = ;

[0111] ;

[0112] in, The number of target pixels after threshold segmentation. This represents the total number of pixels in the block.

[0113] Furthermore, boundary interest scores Represented as:

[0114] ;

[0115] in, The biological boundary enhancement factor is experimentally calibrated to 0.2–0.3. The contribution value to the biological boundary is expressed as:

[0116] ;

[0117] in, The length of the continuous boundary extracted by Canny edge detection. The length of the block's diagonal. For boundary continuity factor, The value range is 0 to 1, for example, 1 for continuous boundaries and 0.3 for discrete boundaries.

[0118] Furthermore, step S303 also includes:

[0119] Based on the magnitude of the interest score above a threshold, the scan priority is assigned to the fine-scanning region; the higher the interest score, the higher the scan priority. For example, highly dynamic regions in living cells that experience sudden vesicle release will be assigned the highest priority due to the sudden increase in the interest score, ensuring priority imaging and avoiding loss of target information due to scan delays.

[0120] Step S304 involves moving the stage 6 to perform super-resolution scanning of the fine scanning area of ​​the biological sample 5 and low-exposure scanning of the coarse scanning area. Typically, the scanning path moves from left to right and from top to bottom of the image to achieve coverage scanning.

[0121] Further, in step S304:

[0122] When the stage 6 performs super-resolution scanning of the fine scan area of ​​the biological sample 5 according to the optimal scanning path, the stage 6 moves with a first step size (small step size), and the image acquisition module 7 acquires super-resolution images with an appropriate exposure of 100 ns to capture the submicron-level distribution of cell membrane proteins. When the stage 6 performs low-exposure scanning of the coarse scan area of ​​the biological sample 5 according to the optimal scanning path, the stage 6 moves with a second step size (large step size), and the image acquisition module 7 acquires low-resolution images with a second exposure, quickly covering areas without critical information; wherein, the first step size is smaller than the second step size, and the first exposure is greater than the second exposure.

[0123] Further, step S4, which involves fusing the super-resolution image and the low-resolution image to obtain the fluorescence image of the biological sample 5 to be tested, includes the following methods:

[0124] Step S401: By matching block coordinates, the super-resolution image and the low-resolution image are stitched together to form a continuous image covering the entire field of view;

[0125] Step S402: Enhance the details of the super-resolution image region in the full-view continuous image and reduce the noise in the low-resolution image region to obtain the fluorescence image of the biological sample 5 to be tested.

[0126] Further, step S402, the method for enhancing the details of the region containing the super-resolution image in the full-view continuous image, includes:

[0127] Based on basic interest score Boundary interest score Build Enhanced Weights , The first coefficient (value taken in this embodiment) ), The second coefficient (value taken in this embodiment) 4), and Range 0~1; The Sobel algorithm is used to search for the structural edge pixels of biological cells in the super-resolution image, and the gray values ​​of the biological cell edge pixels are multiplied by (1+). )% to achieve enhanced details.

[0128] Further, step S402, the method for enhancing the details of the region containing the super-resolution image in the full-view continuous image, includes:

[0129] The fluorescence intensity of all biological cells in the super-resolution image is obtained. The fluorescence intensity of all biological cells is averaged and then multiplied by a preset compensation parameter (0.3 in this embodiment) to obtain a fluorescence enhancement compensation value. The fluorescence enhancement compensation value is superimposed on the area of ​​biological cells in the super-resolution image where the fluorescence intensity of biological cells is less than 200 ADU to achieve detail enhancement.

[0130] Furthermore, step S402, the method for denoising the region containing the low-resolution image, includes:

[0131] The Canny edge detection algorithm is used to identify the regions containing biological cells in the low-resolution image. For pixel blocks between adjacent biological cells, a circular region is constructed based on the minimum straight-line distance between the pixel blocks of adjacent biological cells. A noise reduction weight is used to reduce the noise reduction intensity of this circular region. The noise reduction weight... Represented as:

[0132]

[0133] in, This is a noise reduction constant with a value of 0.8. The noise reduction factor is 0.5. To enhance weight and .

[0134] The low-resolution image is denoised using Gaussian filtering. During the denoising process, the low-resolution image is divided into two parts: the first part is the low-resolution image excluding the annular region, which is then denoised using Gaussian filtering; the second part is the annular region, where the standard deviation of the Gaussian function is replaced with denoising weights. An improved Gaussian filter is obtained, and noise reduction in the annular region is achieved through the improved Gaussian filter.

[0135] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A parallel scanning imaging method based on a planar super-resolution lens array, characterized in that, include: Step S1: Place the biological sample (5) to be tested on the stage (6); Step S2: Control the stage (6) to move the biological sample (5) to be tested, perform a full-field pre-scan of the biological sample (5), and simultaneously acquire images through the image acquisition module (7) while performing the full-field pre-scan of the biological sample (5) to obtain a pre-scan panoramic image. Step S3: Divide the pre-scanned panoramic image into several blocks, and further divide the different blocks into fine-scan regions and coarse-scan regions, ultimately obtaining the super-resolution image corresponding to the fine-scan region and the low-resolution image corresponding to the coarse-scan region. The method includes: Step S301: Divide the pre-scanned panoramic image into n×m blocks, where n is the number of rows in the block and m is the number of columns in the block; Step S302: Calculate the interest score for each block; Step S302 calculates the interest score for each block, expressed as: ； ； in, For the interest score of the block, Based on interest scores, For boundary interest scores, , , The first, second, and third weight coefficients are given, and they satisfy the following conditions: ; The normalized image entropy is expressed as: ； ； ； in, The grayscale information entropy of an image block. grayscale value The probability of occurrence within a block. The maximum possible entropy for 8-bit grayscale; To normalize the texture complexity, it is expressed as: ； ； ； in, For all preset angles of the image block Correspondence Texture complexity obtained by taking the average, for The maximum value in, Regarding the preset angle The complexity, Preset angle And a certain pixel pair under step size The probability of occurrence in all pixel pairs, preset angle For use in determining reference pixels adjacent pixels azimuth angle, Preset angle The number of; The normalized target structure pixel ratio is expressed as: = ； ； in, The number of target pixels after threshold segmentation. This represents the total number of pixels in the block. The boundary interest score Represented as: ； in, The biological boundary enhancement coefficient; The contribution value to the biological boundary is expressed as: ； in, The length of the continuous boundary extracted by Canny edge detection. The length of the block's diagonal. For boundary continuity factors; Step S303: Set a threshold for interest scores, mark blocks with interest scores higher than the threshold as fine scan areas, and mark blocks with interest scores lower than the threshold as coarse scan areas; Step S304: The stage (6) moves to perform super-resolution scanning on the fine scanning area of ​​the biological sample (5) and low-exposure scanning on the coarse scanning area. During the scanning process, the image acquisition module (7) acquires the super-resolution image corresponding to the super-resolution scan and the low-exposure image corresponding to the low-exposure scan. Step S4: Fuse the super-resolution image and the low-resolution image to obtain the fluorescence image of the biological sample (5) to be tested.

2. The parallel scanning imaging system based on a planar super-resolution lens array according to claim 1, characterized in that: Step S303 further includes: Based on the interest score above a threshold, the scan priority is assigned to the fine scan region; the higher the interest score, the higher the scan priority.

3. The parallel scanning imaging method based on a planar super-resolution lens array according to claim 1, characterized in that: In step S304: When the stage (6) performs super-resolution scanning of the fine scanning area of ​​the biological sample (5) according to the optimal scanning path, the stage (6) moves with a step length, and the image acquisition module (7) acquires the super-resolution image with a first exposure. When the stage (6) performs low-exposure scanning of the coarse scanning area of ​​the biological sample (5) according to the optimal scanning path, the stage (6) moves with a second step size, and the image acquisition module (7) acquires a low-resolution image with a second exposure. Wherein, the first step length is less than the second step length, and the first exposure amount is greater than the second exposure amount.

4. The parallel scanning imaging method based on a planar super-resolution lens array according to claim 1, characterized in that: The method for fusing the super-resolution image and the low-resolution image in step S4 to obtain the fluorescence image of the biological sample (5) to be tested includes: Step S401: By matching block coordinates, the super-resolution image and the low-resolution image are stitched together to form a continuous image covering the entire field of view; Step S402: Enhance the details of the super-resolution image in the full-view continuous image and reduce the noise in the low-resolution image to obtain the fluorescence image of the biological sample (5) to be tested.

5. The parallel scanning imaging method based on a planar super-resolution lens array according to claim 4, characterized in that: The method for enhancing details in the region containing the super-resolution image in the full-view continuous image in step S402 includes: Based on basic interest score Boundary interest score Build Enhanced Weights , As the first coefficient, As the second coefficient, and Range 0~1; The Sobel algorithm is used to search for the structural edge pixels of biological cells in the super-resolution image, and the gray values ​​of the biological cell edge pixels are multiplied by (1+). )% to achieve enhanced details.

6. The parallel scanning imaging method based on a planar super-resolution lens array according to claim 4, characterized in that: The method for enhancing details in the region containing the super-resolution image in the full-view continuous image in step S402 includes: The fluorescence intensity of all biological cells in the super-resolution image is obtained. The fluorescence intensity of all biological cells is averaged and then multiplied by a preset compensation parameter to obtain a fluorescence enhancement compensation value. The fluorescence enhancement compensation value is superimposed on the area of ​​biological cells in the super-resolution image where the fluorescence intensity of biological cells is less than 200 ADU to achieve detail enhancement.

7. The parallel scanning imaging method based on a planar super-resolution lens array according to claim 4, characterized in that: The method for noise reduction of the region where the low-resolution image is located in step S402 includes: The Canny edge detection algorithm is used to identify the regions containing biological cells in the low-resolution image. For pixel blocks between adjacent biological cells, a circular region is constructed based on the minimum straight-line distance between the pixel blocks of adjacent biological cells. A noise reduction weight is then used to reduce the noise reduction intensity of this circular region. The noise reduction weight... Represented as: ； in, For noise reduction constant, The noise reduction coefficient is... To enhance weight and , As the first coefficient, The second coefficient; The low-resolution image is denoised using Gaussian filtering. During the denoising process, the low-resolution image is divided into two parts: the first part is the low-resolution image excluding the annular region, which is then denoised using Gaussian filtering; the second part is the annular region, where the standard deviation of the Gaussian function is replaced with denoising weights. An improved Gaussian filter is obtained, and noise reduction in the annular region is achieved through the improved Gaussian filter.

Images