Differential compression structured light illumination super-resolution imaging device based on time sequence multiplexing

Abstract

The invention discloses a differential compression structured light illumination super-resolution imaging device based on time sequence multiplexing. The differential compression structured light illumination super-resolution imaging device is composed of a light source and beam expanding system, a structured light illumination imaging system, a coding aperture time domain compression imaging system, a time sequence control system and a data processing reconstruction system. According to the invention, the two digital micromirror devices are controlled through a time sequence, and wide-field and structured light compression imaging is realized respectively. In the wide-field mode, the two digital micromirror devices are switched to the total transmission mode, and fluorescence signals of the excitation sample are directly collected by the camera; in a structured light mode, the first digital micro-mirror device generates six time-varying stripe coding patterns, and dynamic space coding of the second digital micro-mirror device is combined, so that a camera obtains a compressed image through single exposure. Through differential processing of a wide field and a compressed image, background noise is suppressed, high-frequency information weight is enhanced, the bottleneck of high-frequency signal offset in traditional compressed imaging is broken through, image reconstruction quality is remarkably improved, and high-fidelity imaging of a high-speed dynamic scene is realized.

Description

Technical Field

[0001] The present invention relates to the fields of structured light illumination super-resolution imaging, compressive sensing and differential imaging technologies, and specifically to a differential compressive structured light illumination super-resolution imaging device based on temporal multiplexing. Aiming at the problem in traditional compressive imaging-based structured light illumination imaging technology that the superposition compression of structured light illumination images with different phase shifts suppresses the proportion of high-frequency components in spectral modulation, the present invention uses the differential processing of compressive images and wide-field images to improve the proportion of high-frequency information components in the image while enhancing the signal-to-noise ratio, thereby significantly improving the quality of super-resolution image reconstruction. At the same time, through the temporal multiplexing design, the structure of the experimental device is simplified, reducing the complexity of the device. In addition, the present invention can be extended to fields such as fluorescence correlation spectroscopy and in vivo tissue imaging, providing high-quality and high-precision imaging tools for biomedical researchers to analyze ultra-high-speed processes such as neuronal synaptic transmission and mitochondrial dynamic interactions. Background Art

[0002] As a core tool for exploring the microscopic world, the spatial resolution of optical imaging technology has long been limited by the optical diffraction limit (about 200 nm), making it difficult to resolve the fine dynamics of subcellular structures (such as mitochondrial cristae, nuclear pore complexes, etc.). Although super-resolution imaging technologies such as stimulated emission depletion microscopy (STED) and single molecule localization microscopy (PALM / STORM) have broken through this limitation, their inherent defects limit their wide application: STED relies on high-intensity laser scanning, which is prone to cause serious phototoxicity and limits its application in the dynamic observation of living cells; PALM / STORM needs to accumulate thousands of frames of sparse luminescence images, and the imaging speed is as low as the minute level, unable to capture millisecond-level biological events. In contrast, structured illumination microscopy (SIM) shifts high-frequency information to the low-frequency detection range through multi-directional structured light modulation, and only 9 frames of original images are required to reconstruct the super-resolution result. It has the advantages of high imaging speed and low light damage, and has become an important tool for the dynamic research of living cells.

[0003] However, traditional SIM technology still faces multiple challenges. First, its super-resolution reconstruction relies on the accurate extraction of high-frequency information. In actual imaging, a large number of images need to be acquired at multiple angles and phases, which exacerbates the problems of photobleaching and phototoxicity and makes it difficult to meet the needs of long-term in vivo observation. Second, the SIM reconstruction algorithm is sensitive to noise, and light distortion, parameter estimation errors, or insufficient sampling will introduce artifacts and reduce the imaging quality. In addition, the temporal resolution of traditional SIM is limited by the camera frame rate and mode switching efficiency, making it difficult to meet the observation needs of ultra-high-speed dynamic processes (such as the opening and closing of ion channels and the interaction of organelles). In recent years, compressive sensing-based SIM (CISIM) has significantly improved the imaging speed by encoding and compressing techniques to fuse multi-frame information into a single-frame acquisition. However, the superposition and compression of multiple frames will cause the cancellation of high-frequency signals, thus suppressing the proportion of high-frequency modulation signals and resulting in a decrease in the reconstructed resolution. Therefore, there is an urgent need for a technical solution that can efficiently retain high-frequency modulation signals in compressive imaging to significantly improve the reconstruction quality of super-resolution images. Summary of the Invention

[0004] The object of the present invention is to address the problems in existing compressive imaging-based structured light illumination imaging technologies, such as the easy cancellation of high-frequency signals and the difficulty of image reconstruction. A differential compression structured light illumination super-resolution imaging device based on time-division multiplexing is proposed. By realizing the efficient acquisition of wide-field images and compressed images, and enhancing the proportion of high-frequency information through differential compression images and wide-field images, high-quality imaging of images that cannot be achieved by conventional compressive imaging-based structured light illumination imaging is achieved.

[0005] The specific technical solution to achieve the object of the present invention is as follows:

[0006] A structured light illumination super-resolution imaging device based on differential compressive sensing, which includes:

[0007] A light source and beam expander system composed of a continuous laser and a beam expander;

[0008] The continuous laser and the beam expander of the light source and beam expander system are connected in sequence in the optical path;

[0009] A structured light illumination imaging system composed of a first digital micromirror device, a linear polarizer, a first lens, a spatial filter, a partitioned half-wave plate, a second lens, a third lens, a dichroic mirror, an objective lens, and a sample stage;

[0010] The first digital micromirror device, the linear polarizer, the first lens, the spatial filter, the partitioned half-wave plate, the second lens, the third lens, the dichroic mirror, the objective lens, and the sample stage of the structured light illumination imaging system are connected in sequence in the optical path;

[0011] An encoded aperture time-domain compressive imaging system composed of a relay lens, a beam splitter cube, a fourth lens, a fifth lens, a second digital micromirror device, and a CMOS camera;

[0012] The relay lens and the beam splitter cube of the coded aperture time-domain compressed imaging system are optically connected in sequence; one path transmitted by the beam splitter cube is optically connected to the fourth lens, the fifth lens, and the second digital micromirror device in sequence; one path reflected by the beam splitter cube is optically connected to the CMOS camera;

[0013] A synchronization control system composed of a field programmable gate array FPGA;

[0014] A data processing and reconstruction system composed of a computer;

[0015] The light source, the beam expander of the beam expansion system, and the first digital micromirror device of the structured light illumination imaging system are optically connected;

[0016] One path reflected by the dichroic mirror of the structured light illumination imaging system is optically connected to the relay lens;

[0017] The field programmable gate array FPGA of the synchronization control system is respectively connected to the first digital micromirror device of the structured light illumination imaging system, the second digital micromirror device of the coded aperture time-domain compressed imaging system, and the CMOS camera data line;

[0018] The computer of the data processing and reconstruction system is respectively connected to the CMOS camera of the coded aperture time-domain compressed imaging system and the field programmable gate array FPGA data line of the synchronization control system.

[0019] Furthermore, the light source and the beam expansion system generate laser light and expand it. In the structured light illumination mode, the field programmable gate array FPGA of the synchronization control system triggers the first digital micromirror device of the structured light illumination imaging system in a time-sharing manner, enabling it to generate three different directions in sequence, with a phase difference of 2π / 3 applied to each direction for two-step phase shift, resulting in a total of six structured light patterns; after the laser polarization state is modulated in cooperation with the linear polarizer and the partitioned half-wave plate, the first lens and the spatial filter are used to filter out stray light and homogenize the light field, and then it is relayed to the objective lens through the second lens and the third lens to excite the fluorescence signal of the sample to be measured on the sample stage; the fluorescence signal is reflected by the dichroic mirror and enters the coded aperture time-domain compressed imaging system, and is projected onto the second digital micromirror device through the transmission optical path of the beam splitter cube; the field programmable gate array FPGA of the synchronization control system synchronously controls the second digital micromirror device to load a dynamic pseudo-random coding template, performs time-domain compression modulation on the fluorescence images corresponding to the six structured light patterns, superimposes the six images into a single-frame compressed signal through spatial coding, and then performs single-exposure acquisition by the CMOS camera.

[0020] Further, the light source and the beam expander system generate laser light and expand it. In the wide-field mode, the field-programmable gate array (FPGA) of the synchronization control system controls the first digital micromirror device in the structured light illumination imaging system and the second digital micromirror device in the coded aperture time-domain compressed imaging system to switch to the full-transmission mode. In the structured light illumination imaging system, the light without structured light modulation directly hits the sample stage, exciting the fluorescence signal of the sample to be measured on the sample stage. The fluorescence signal is reflected by the dichroic mirror and enters the coded aperture time-domain compressed imaging system, where it is projected onto the second digital micromirror device through the transmission optical path of the beam splitter cube and then directly returns to the beam splitter cube without coding and compression, and is directly projected onto the CMOS camera by the reflection of the beam splitter cube. At the same time, the FPGA of the synchronization control system triggers the CMOS camera in a time-sharing manner, and an unloaded structured light wide-field image is independently acquired after the structured light mode acquisition for subsequent processing by the differential algorithm.

[0021] Further, the synchronization control system realizes the timing switching between the wide-field imaging mode and the structured light illumination mode, the dynamic coding template loading, and the efficient synchronous acquisition of the compressed image and the wide-field image by time-sharing regulation of the trigger signals of the first digital micromirror device, the second digital micromirror device, and the CMOS camera.

[0022] Further, the data processing and reconstruction system executes the frequency-domain differential algorithm to suppress the low-frequency background noise and enhance the proportion of the high-frequency modulation signal through the spectral differential operation of the compressed image and the wide-field image. Subsequently, the ADMM-DIP algorithm, a hybrid optimization framework of the deep image prior (DIP) network and the alternating direction method of multipliers (ADMM), is used to iteratively reconstruct the differential image. The compressed sensing coding matrix and the observed data are used as inputs to iteratively reconstruct the super-resolution image ; an optimization objective function is constructed:

[0023] In the above formula, is the prior image generated by the deep image prior network, and are the trade-off parameters; the alternating direction method of multipliers performs the following iterative optimization:

[0024]

[0025] In the above formula, is the number of alternating iterations of the optimization framework, fixing the network parameter , solving the linear inverse problem by the conjugate gradient method, and updating the image ;

[0026]

[0027] In the above formula, the fixed image , optimize the network parameters through backpropagation , enhance high-frequency details and structural continuity; then introduce the regularization term L1 to suppress noise and enhance sparse constraints; after 50 to 100 iterations, output a super-resolution image that fuses differential enhancement of high-frequency information and sparse optimization.

[0028] Finally, fuse the high-frequency information enhanced by differential enhancement and the joint optimization strategy, and the device outputs a reconstructed image with a resolution reaching the subcellular level, significantly improving the visualization effect of subcellular structures such as mitochondrial cristae and nuclear pore complexes.

[0029] The present invention includes optical sampling and an image reconstruction algorithm. Optical sampling is responsible for the high-speed compressed acquisition of high-resolution information of dynamic targets, while the image reconstruction algorithm realizes the high-quality restoration of super-resolution images through differential enhancement and sparse optimization.

[0030] The present invention has the following technical advantages:

[0031] The present invention compresses the image and the wide-field image through differential processing, effectively retains the high-frequency modulation signal, significantly improves the imaging quality, and is particularly suitable for high-resolution imaging requirements under low-light conditions. Compared with the traditional three-step phase-shift structured light illumination scheme, the compressive imaging of the present invention adopts an encoding strategy of two-step phase shift with a phase difference of 2π / 3 applied in each of three directions, reducing the signal interference caused by the superposition of multiple frames of images. While improving the signal-to-noise ratio, the proportion of high-frequency information increases, thereby achieving high-quality image imaging.

[0032] Through the design of a time-division multiplexing architecture, the present invention integrates the image acquisition processes of the wide-field mode and the structured light mode into the same optical path, reduces redundant optical elements, and significantly reduces the device complexity. This design not only simplifies the device structure but also realizes mode switching and signal synchronization through precise timing control of the field-programmable gate array FPGA, improving the stability and reliability of the device.

[0033] The present invention innovatively integrates differential enhancement and compressive sensing technologies, and reconstructs super-resolution images from single-frame compressed data through the ADMM-DIP framework. Among them, the ADMM-DIP framework extracts multi-scale features through a deep image prior network, combines alternating iterative optimization, and under the joint action of sparse constraints and data fidelity terms, improves the reconstruction resolution to the subcellular level, meeting the high-precision imaging requirements for organelle morphology, protein distribution, etc. in the biomedical field.

[0034] The present invention can support the real-time observation of ultra-high-speed dynamic processes, and is particularly suitable for observing ultra-high-speed biological events such as neuron firing and organelle interactions. Description of the Drawings

[0035] Figure 1Schematic diagram of the structure of the present invention;

[0036] Figure 2 Flow chart of the operation of the present invention;

[0037] Figure 3 Original, wide-field, compressed reconstruction, and time-division multiplexed differential compression reconstruction imaging results of biological sample super-resolution imaging using the present invention;

[0038] Figure 4 Comparison chart of the reconstruction quality of time-division multiplexed differential compression reconstruction images (TDCISIM) at different noise levels;

[0039] Figure 1 In the figure: 100 - light source and beam expander system; 101 - continuous laser; 102 - beam expander; 200 - structured light illumination imaging system; 201 - first digital micromirror device; 202 - linear polarizer; 203 - first lens; 204 - spatial filter; 205 - segmented half-wave plate; 206 - second lens; 207 - third lens; 208 - dichroic mirror; 209 - objective lens; 210 - sample stage; 300 - coded aperture time-domain compressive imaging system; 301 - relay lens; 302 - beam splitter cube; 303 - fourth lens; 304 - fifth lens; 305 - second digital micromirror device; 306 - CMOS camera; 400 - synchronization control system; 401 - field programmable gate array FPGA; 500 - data processing and reconstruction system; 501 - computer. Detailed implementation manner

[0040] Refer to Figure 1 , the present invention is a time-division multiplexed differential compression structured light illumination super-resolution imaging device, including: a light source and beam expander system 100 composed of a continuous laser 101 and a beam expander 102;

[0041] The continuous laser 101 and the beam expander 102 of the light source and beam expander system 100 are connected in optical path in sequence;

[0042] A structured light illumination imaging system 200 composed of a first digital micromirror device 201, a linear polarizer 202, a first lens 203, a spatial filter 204, a segmented half-wave plate 205, a second lens 206, a third lens 207, a dichroic mirror 208, an objective lens 209, and a sample stage 210;

[0043] The first digital micromirror device 201, the linear polarizer 202, the first lens 203, the spatial filter 204, the segmented half-wave plate 205, the second lens 206, the third lens 207, the dichroic mirror 208, the objective lens 209, and the sample stage 210 of the structured light illumination imaging system 200 are connected in optical path in sequence;

[0044] An encoded aperture time-domain compressed imaging system 300 composed of a relay lens 301, a beam-splitting cube 302, a fourth lens 303, a fifth lens 304, a second digital micromirror device 305, and a CMOS camera 306;

[0045] The relay lens 301 and the beam-splitting cube 302 of the encoded aperture time-domain compressed imaging system 300 are connected in optical path in sequence; one path transmitted by the beam-splitting cube 302 is connected in optical path with the fourth lens 303, the fifth lens 304, and the second digital micromirror device 305 in sequence; one path reflected by the beam-splitting cube 302 is connected in optical path with the CMOS camera 306;

[0046] A synchronization control system 400 composed of a field programmable gate array FPGA 401;

[0047] A data processing and reconstruction system 500 composed of a computer 501;

[0048] The beam expander 102 of the light source and beam expander system 100 is connected in optical path with the first digital micromirror device 201 of the structured light illumination imaging system 200;

[0049] One path reflected by the dichroic mirror 208 of the structured light illumination imaging system 200 is connected in optical path with the relay lens 301;

[0050] The field programmable gate array FPGA 401 of the synchronization control system 400 is respectively connected with the first digital micromirror device 201 of the structured light illumination imaging system 200, the second digital micromirror device 305 of the encoded aperture time-domain compressed imaging system 300, and the CMOS camera 306 by data lines;

[0051] The computer 501 of the data processing and reconstruction system 500 is respectively connected with the CMOS camera 306 of the encoded aperture time-domain compressed imaging system 300 and the field programmable gate array FPGA 401 of the synchronization control system 400 by data lines.

[0052] The present invention works as follows:

[0053] Refer to Figure 1 and Figure 2, the continuous laser 101 in the light source and beam expansion system 100 emits a laser beam, which is collimated and expanded by the beam expander 102 and then transmitted to the structured light illumination imaging system 200. In the structured light mode, the first digital micromirror device 201 generates three directions under the timing control of the field programmable array FPGA 401 of the synchronous control system 400, and applies a two-step phase shift with a phase difference of 2π / 3 in each direction, for a total of six structured light modes 4. The linear polarizer 202 and the partitioned half-wave plate 205 coordinately modulate the polarization state of the laser to ensure that the contrast of the structured light pattern under different phase shifts is maximized; after the spatial filter 204 filters out the non-uniform components of the light field, it is relayed to the objective lens 209 by the lens group 206 and 207, accurately stimulating the fluorescence signal of the sample to be tested on the sample stage 210, and generating a structured light modulated image.

[0054] See also Figure 1 and Figure 2 The fluorescence signal generated by the excitation is reflected by the dichroic mirror 208 and enters the coded aperture time domain compression imaging system 300. After passing through the beam splitting cube 302 transmission light path, the fourth lens 303 and the fifth lens 304 relay the fluorescence image to the second digital micromirror device 305. Under the control of the field programmable array FPGA401, the second digital micromirror device dynamically loads the pseudo-random coding template, performs time domain compression coding on the six frames of structured light images, and finally the CMOS camera 306 obtains the compressed image by a single exposure.

[0055] See also Figure 1 and Figure 2 , the continuous laser 101 in the light source and beam expansion system 100 emits a laser beam, which is collimated and expanded by the beam expander 102 and then transmitted to the structured light illumination imaging system 200. In the wide field mode, the field programmable array FPGA 401 controls the first digital micromirror device and the second digital micromirror device to switch to the full transmission mode, and the fluorescence signal of the excited sample is directly transmitted to the CMOS camera through the reflection light path of the beam splitter cube 302, so as to realize the collection of unmodulated wide field images as the reference data for the subsequent frequency domain difference processing.

[0056] See also Figure 1, the synchronous control system 400 coordinates the operation timing of the first digital micromirror device, the second digital micromirror device and the CMOS camera 306 through the trigger signal: the first digital micromirror device switches six structured light modes according to the preset sequence, the second digital micromirror device synchronously loads the corresponding coding template, and the CMOS camera collects the compressed image and the wide field image in time-sharing. After the acquisition is completed, the computer 501 of the data processing and reconstruction system 500 executes the differential algorithm to increase the proportion of high-frequency information. Subsequently, the ADMM-DIP network optimization framework is used for iterative reconstruction: among them, the deep image prior network DIP adaptively captures the spatial correlation and contour details of the subcellular structure through the multi-scale feature extraction module integrating the multi-head self-attention mechanism; the alternating direction multiplier method ADMM suppresses noise interference while retaining high-frequency information by alternately solving the data fidelity term and the sparse regularization term. Finally, after 50 to 100 rounds of alternating optimization iterations, the system outputs a super-resolution image with both high resolution and high signal-to-noise ratio, which can clearly analyze the dynamic details of subcellular structures such as mitochondrial cristae and nuclear pore complexes.

[0057] See also Figure 1 The present invention integrates wide field and structured light modes into the same optical path through a time-sequential multiplexing architecture, simplifies the experimental optical path diagram, and greatly reduces the complexity of the experimental imaging device. Experimental verification shows that the device increases the proportion of high-frequency image information in the observation of interactions between living organelles, improves the quality of reconstructed images, and improves the signal-to-noise ratio compared to traditional CISIM, providing a high-fidelity imaging tool for ultra-high-speed dynamic processes such as neuronal discharge and vesicle transport.

[0058] Example 1

[0059] In order to verify the super-resolution advantage of the present invention over traditional CISIM, three typical biological samples, namely, microtubule-binding protein of embryonic rat hippocampal neurons, microtubule of COS-7 cells and Trio protein of Swiss 3T3 cells, were selected for simulation and reconstruction under the conditions of wavelength of 550 nm, objective lens NA 1.5 and pixel size of 30 nm.

[0060] See also Figure 3 , using the device of the present invention, images of living biological samples are selected for simulation and reconstruction results. (a) and (f) are original images; (b) and (g) are wide-field measurement images; (c) and (h) are CISIM reconstruction images; (d) and (i) are TDCISIM reconstruction images. They are images of microtubule-binding proteins in embryonic rat hippocampal neurons and microtubules in COS-7 cells, respectively, and intensity curves of selected areas with markers (e) and (j).

[0061] See also Figure 3, after the original image is modulated by structured light, measurement data is generated through the encoding and compression models of CISIM and the TDCISIM of the present invention respectively, and super-resolution reconstruction is performed using corresponding algorithms. The experimental results show that the microtubule edge sharpness and the spatial distribution of nuclear pore complexes in the reconstructed images of TDCISIM ( Figure 3 d, i) are closer to the original image, significantly superior to CISIM ( Figure 3 c, h). By extracting the intensity distribution of the marked line, it can be seen that TDCISIM has a higher reduction degree of the intensity fluctuation of the 30-nm-wide microtubule structure and an obvious artifact suppression effect.

[0062] Refer to Figure 3 , the simulation reconstruction based on the experimental data of the BioSR database shows that TDCISIM can still resolve the mitochondrial crista substructure in the single-frame compression mode. Although its resolution is slightly lower than that of traditional SIM, this is because additional encoding and compression operations are adopted in TDCISIM, which inevitably leads to some information loss during the sampling process. However, the imaging speed is improved compared with SIM, and the phototoxicity is greatly reduced, making it more suitable for long-term dynamic observation of living cells.

[0063] Example 2

[0064] To evaluate the robustness of the present invention in a complex noise environment, a super-resolution reconstruction experiment under Gaussian white noise interference is designed. First, a simulation image containing a 30-nm-wide curve array is generated, which is convolved with a point spread function with a full width at half maximum of 60 nm as the original target. The TDCISIM differential image is generated through structured light modulation and time-domain compression encoding, and Gaussian white noise with noise intensities σ of 0, 0.01, 0.02, and 0.03 is injected respectively.

[0065] Refer to Figure 4 , which shows the comparison of the reconstruction quality of TDCISIM under different noise levels. (a) is the original image; (b) is the wide-field image, which shows the diffraction-limited low-frequency characteristics after convolution with the point spread function; (c), (e), (f), (g) are the reconstructed images of TDCISIM with added Gaussian white noise intensities σ = 0, 0.01, 0.02, and 0.03 respectively; (d) shows the intensity distribution curves at the solid line marked positions of different images; (h) is the curve of the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) of the reconstructed images versus the noise intensity σ.

[0066] Refer to Figure 4, under the condition of no noise (σ = 0, Figure c), the reconstructed image clearly restores the original curve contour; as the noise increases (Figures e, f, g), the edges of the image appear slightly blurred, but the key structures can still be identified; especially under the condition of high noise (σ = 0.03, Figure g), the TDCISIM reconstructed image can still clearly distinguish the curve contour, and the artifact intensity is significantly lower than that of the wide-field image (Figure b); through the intensity distribution curve (Figure d), it can be seen that the TDCISIM reconstruction result is highly consistent with the original target, while the wide-field image cannot distinguish details due to insufficient resolution.

[0067] See Figure 4 , through local magnification and comparison of intensity distribution curves, it can be seen that the increase in noise level causes slight blurring of the edges of the reconstructed image, but the key structures can still be identified. Quantitative analysis shows that PSNR and SSIM show a linear and gradual decline trend with the increase in noise, but the system has strong anti-noise ability, verifying its robustness in complex noise environments. Experiments prove that the present invention can stably output subcellular-resolution images under certain conditions, providing high-fidelity imaging guarantee for low-light or high-speed dynamic scenarios (such as neuronal synaptic transmission).

[0068] The above Examples 1 and 2 fully verify the core advantages of the present invention: through the cooperation of the differential compressive sensing architecture and the deep optimization algorithm, while improving the imaging speed, the weight of high-frequency signals and the noise robustness are significantly enhanced. Compared with the prior art, the present invention not only solves the bottleneck of high-frequency information cancellation in compressive imaging, but also reduces the system complexity through the time-division multiplexing design, providing an innovative tool for the real-time observation of ultra-high-speed dynamic processes in biomedicine.

Claims

1. A differential compression structured light illumination super-resolution imaging device based on time-series multiplexing, characterized in that: It includes: A light source and beam expansion system (100) consisting of a continuous laser (101) and a beam expander (102); The light source is optically connected in sequence to a continuous laser (101) and a beam expander (102) of a beam expansion system (100); A structured light illumination imaging system (200) composed of a first digital micromirror device (201), a linear polarizer (202), a first lens (203), a spatial filter (204), a partitioned half-wave plate (205), a second lens (206), a third lens (207), a dichroic mirror (208), an objective lens (209), and a sample stage (210); The first digital micromirror device (201), the linear polarizer (202), the first lens (203), the spatial filter (204), the partitioned half-wave plate (205), the second lens (206), the third lens (207), the dichroic mirror (208), the objective lens (209) and the sample stage (210) of the structured light illumination imaging system (200) are optically connected in sequence; A coded aperture time-domain compression imaging system (300) composed of a relay lens (301), a beam splitting cube (302), a fourth lens (303), a fifth lens (304), a second digital micromirror device (305), and a CMOS camera (306); The relay lens (301) and the beam splitter cube (302) of the coded aperture time domain compression imaging system (300) are optically connected in sequence; a transmission path of the beam splitter cube (302) is optically connected in sequence to a fourth lens (303), a fifth lens (304), and a second digital micromirror device (305); and a reflection path of the beam splitter cube (302) is optically connected to a CMOS camera (306); A synchronous control system (400) composed of a field programmable array FPGA (401); A data processing and reconstruction system (500) composed of a computer (501); The light source and beam expander (102) of the beam expander system (100) are optically connected to the first digital micromirror device (201) of the structured light illumination imaging system (200); A dichroic mirror (208) of the structured light illumination imaging system (200) reflects one path and is connected to the optical path of the relay lens (301); The field programmable array FPGA (401) of the synchronous control system (400) is respectively connected to the first digital micromirror device (201) of the structured light illumination imaging system (200), the second digital micromirror device (305) of the coded aperture time domain compression imaging system (300), and the data line of the CMOS camera (306); The computer (501) of the data processing and reconstruction system (500) is respectively connected to the CMOS camera (306) of the coded aperture time domain compression imaging system (300) and the field programmable array FPGA (401) of the synchronous control system (400) through data lines; wherein: The data processing and reconstruction system (500) executes a frequency domain difference algorithm to suppress low-frequency background noise and enhance the proportion of high-frequency modulation signals by performing a frequency spectrum difference operation between a compressed image and a wide-field image; then, an ADMM-DIP algorithm, a hybrid optimization framework of a deep image prior network DIP and an alternating direction multiplier method ADMM, is used to iteratively reconstruct the differenced image; and the compressed sensing coding matrix With observational data Iteratively reconstruct the super-resolution image as input ; Construct optimization objective function: ; In the above formula, The prior image generated by the deep image prior network, and To trade off the parameters, the alternating direction multiplier method performs the following iterative optimization: ； In the above formula, To optimize the number of alternating iterations of the framework, the network parameters are fixed , solve the linear inverse problem by the conjugate gradient method and update the image ; ； In the above formula, the fixed image , optimize network parameters through back propagation , enhancing high-frequency details and structural continuity; then the regularization term L1 is introduced to suppress noise and enhance sparse constraints; after 50 to 100 rounds of iterations, the super-resolution image that fuses differential enhancement of high-frequency information and sparse optimization is output.

2. The differential compression structured light illumination super-resolution imaging device based on time-series multiplexing according to claim 1, characterized in that: The light source and beam expansion system (100) generates laser light and expands the beam thereof. In a structured light illumination mode, the field programmable array FPGA (401) of the synchronous control system (400) is used to time-share trigger the first digital micromirror device (201) of the structured light illumination imaging system (200), so that it sequentially generates three different directions, each direction applying a two-step phase shift with a phase difference of 2π / 3, for a total of six structured light modes; after the linear polarizer (202) and the partitioned half-wave plate (205) coordinately modulate the polarization state of the laser light, the first lens (203) and the spatial filter (204) filter out stray light and homogenize the light field, and then the second lens (206) and the third lens (207) are used to transmit the laser light to the optical system. The fluorescence signal is then transmitted to the objective lens (209) to excite the fluorescence signal of the sample to be tested on the sample stage (210); the fluorescence signal is reflected by the dichroic mirror (208) and enters the coded aperture time domain compression imaging system (300), and is projected to the second digital micromirror device (305) through the transmission light path of the beam splitting cube (302); the field programmable array FPGA (401) of the synchronous control system (400) synchronously controls the second digital micromirror device (305) to load a dynamic pseudo-random coding template, performs time domain compression modulation on the fluorescence images corresponding to the six structured light modes, and superimposes the six images into a single frame compression signal through spatial coding, and then performs single exposure acquisition through the CMOS camera (306).

3. The differential compression structured light illumination super-resolution imaging device based on time-series multiplexing according to claim 1, characterized in that: The light source and beam expansion system (100) generates laser light and expands the beam thereof. In a wide-field mode, a field programmable array FPGA (401) of a synchronous control system (400) controls a first digital micromirror device (201) in a structured light illumination imaging system (200) and a second digital micromirror device (305) in a coded aperture time-domain compression imaging system (300) to switch to a full-transmission mode. In the structured light illumination imaging system (200), light without structured light modulation directly hits a sample stage (210) to excite a fluorescence signal of a sample to be measured on the sample stage (210). The fluorescence signal After being reflected by the dichroic mirror (208), the light enters the coded aperture time domain compression imaging system (300), is projected to the second digital micromirror device (305) through the transmission light path of the beam splitting cube (302), and then directly returns to the beam splitting cube (302) without being coded and compressed. The light is directly projected to the CMOS camera (306) by reflection from the beam splitting cube (302), and the CMOS camera (306) is simultaneously triggered by the field programmable array FPGA (401) of the synchronous control system (400) in a time-sharing manner. After the structured light mode is acquired, a wide-field image without structured light is independently acquired for subsequent processing by a differential algorithm.

4. The differential compression structured light illumination super-resolution imaging device based on time-series multiplexing according to claim 1, characterized in that: The synchronous control system (400) achieves the time-sequential switching of the wide-field imaging mode and the structured light illumination mode, the dynamic coding template loading, and the efficient synchronous acquisition of the compressed image and the wide-field image by time-sharingly regulating the trigger signals of the first digital micromirror device (201), the second digital micromirror device (305), and the CMOS camera (306).

Images