Ultrahigh-speed imaging device based on acousto-optic filtering modulation

Abstract

The invention discloses an ultra-high-speed imaging device based on acousto-optic filtering modulation. The ultra-high-speed imaging device comprises a spectrum modulation system, a spectrum imaging system and a data processing system. The spectrum modulation system utilizes an acousto-optic tunable filter (AOTF) to dynamically modulate broadband continuous spectrum light, narrow-band illumination light which rapidly changes along with time is generated, and time information of a dynamic scene is effectively mapped to a spectrum dimension. And the spectral imaging system images the spectral coded dynamic scene to the hyperspectral camera to realize acquisition of spectral space-time information. The data processing system performs spectral crosstalk correction, channel separation and demosaicing reconstruction on the acquired original mosaic image, and recovers a time sequence image according to a spectrum-time mapping relation, thereby realizing single-shot superspeed imaging in a dynamic process. The device has the advantages of being simple and compact in structure, flexible in time window, high in imaging fidelity and the like, and has wide application prospects in the fields of rapid dynamics such as microfluidics, laser processing and ultrafast physics.

Description

Technical Field

[0001] This invention relates to the field of optical imaging technology, and more specifically to an ultra-high-speed imaging device based on acousto-optic filtering modulation. Background Technology

[0002] Ultra-high-speed imaging technology has significant applications in scientific research, industrial inspection, and biomedicine, enabling the capture of the dynamic evolution of various transient processes, such as laser-material interactions, plasma dynamics, chemical reaction kinetics, and biological neural signal transduction. However, traditional ultra-high-speed imaging methods, such as rotating mirror cameras, framing cameras, and high-speed sampling cameras, typically rely on complex optical components or the simultaneous operation of multiple detectors to achieve high frame rates. These methods suffer from problems such as complex system structure, high cost, limited imaging field of view, and insufficient scalability.

[0003] To overcome these limitations, researchers have proposed high-speed imaging methods based on active illumination. The core idea is to map temporal information to other physical domains, such as polarization, wavelength, angle, and frequency, to indirectly capture dynamic scenes. Among these, ultra-high-speed imaging methods based on time-wavelength mapping have attracted widespread attention because they can directly map the temporal information of dynamic scenes to the spectral domain and efficiently reconstruct time series through spectral measurements. These methods typically use pulses or pulse sequences with time-varying spectral components as the illumination source, encoding the temporal evolution of the dynamic process into the spectral signal. Subsequently, spectral information is converted into a continuous time series image using spectral resolution detection. Typical examples include sequential timing full optical mapping photography (STAMP), compressed ultrafast spectral time photography (CUST), and chirped spectral mapping ultrafast imaging (CSMUP). These methods have achieved significant progress in imaging speed and sequence depth, but they generally rely on ultrafast laser sources, resulting in high system cost, large size, and insufficient flexibility in the imaging window. Furthermore, the image fidelity is reduced during the compression reconstruction process. Summary of the Invention

[0004] The purpose of this invention is to provide an ultra-high-speed imaging device based on acousto-optic filtering modulation, addressing the shortcomings of existing technologies. This device utilizes a broadband supercontinuum light source combined with an acousto-optic tunable filter (AOTF). By dynamically controlling the diffraction wavelength of the AOTF through time-varying radio frequency signals, it generates a narrowband illumination light that changes rapidly over time, mapping the temporal evolution of a dynamic scene to the spectral domain. Subsequently, a hyperspectral camera acquires the spectrally encoded dynamic scene and obtains image information for each spectral channel through spectral channel separation. Ultimately, high-fidelity, high-resolution single-exposure ultra-high-speed imaging is achieved, providing a reliable technical means for real-time observation of complex transient processes.

[0005] The specific technical solution for achieving the objective of this invention is as follows: An ultra-high-speed imaging device based on acousto-optic filtering modulation, comprising: A spectral modulation system consisting of a supercontinuum light source, a short-pass filter, a linear polarizer, an acousto-optic tunable filter (AOTF), and a first lens; The supercontinuum light source, short-wave pass filter, linear polarizer, acousto-optic tunable filter (AOTF), and first lens of the spectral modulation system are connected in sequence via optical paths. A spectral imaging system consisting of a sample to be tested, an objective lens, a second lens, and a hyperspectral camera; The sample to be tested, objective lens, second lens and hyperspectral camera of the spectral imaging system are connected in sequence via optical paths. A data processing system composed of computers; The first lens of the spectral modulation system is connected to the optical path of the sample under test in the spectral imaging system. The computer of the data processing system is connected to the acousto-optic tunable filter (AOTF) of the spectral modulation system and the data cable of the hyperspectral camera of the spectral imaging system. The supercontinuum light source outputs broadband supercontinuum light with a spectral coverage of 410nm-2400nm, a pulse width of approximately 6ps, a maximum power of 7W, and a repetition frequency adjustable between 10kHz and 80MHz. The broadband supercontinuum light is filtered by a short-pass filter with a cutoff wavelength of 600nm, allowing only light with wavelengths less than 600nm to pass through, thus reducing the spectral range to 410nm-600nm. It then passes through a linear polarizer, and by adjusting its transmission direction, the transmitted broadband light is made to be linearly polarized with its polarization direction along the vertical direction. Subsequently, broadband light with a spectral range of 410nm-600nm and vertical linear polarization enters the acousto-optic tunable filter (AOTF). The AOTF operates in continuous frequency sweep mode, performing high-speed time-sequential scanning filtering on the incident broadband light by setting the frequency scan range and total scan time of the radio frequency signal. This results in the sequential output of narrowband light with different center wavelengths along the time axis, forming illumination light that changes rapidly over time. In this invention, the scanning band is set to 450nm-600nm, and the total scan time is flexibly adjusted according to different experimental scenarios. The time-varying illumination light with a spectral range of 450nm-600nm emitted from the AOTF is focused by the first lens and then illuminates the surface of the sample under test. The sample can be a dynamically moving object or a stationary object; if the sample moves, it constitutes a dynamic scene. Under the illumination of the time-varying illumination light, the time information of the dynamic scene, which evolves over time, is converted into the spectral dimension through a time-spectral mapping relationship, thus generating a spectrally encoded dynamic scene with two-dimensional spatial and spectral information. The spectrally encoded dynamic scene is magnified and imaged onto the photosensitive surface of the hyperspectral camera via an objective lens and a second lens. The magnification can be adjusted according to experimental requirements and sample size. The hyperspectral camera acquires the complete spectrally encoded dynamic scene in a single exposure, obtaining the raw mosaic image, which is then transmitted to the computer in the data processing system. Finally, the computer sequentially performs three algorithmic steps on the acquired raw mosaic image: spectral crosstalk correction, channel separation, and demosaic reconstruction. Based on the spectral-time mapping relationship, it accurately reconstructs the time-series image of the dynamic scene, thereby achieving ultra-high-speed imaging of the dynamic process in a single shot.

[0006] Furthermore, the spectral modulation system dynamically modulates broadband light through acousto-optic filtering to generate illumination light that varies with time, and projects this illumination light onto the surface of the sample to be tested, thereby mapping the temporal evolution information of the sample to the spectral domain.

[0007] Furthermore, the spectral imaging system acquires the illumination light that passes through the sample under test and images it onto the hyperspectral camera to obtain the spectral encoding spatiotemporal information of the sample under test.

[0008] Furthermore, the data processing system processes the original mosaic image acquired by the hyperspectral camera. First, it suppresses crosstalk between spectral channels based on a preset spectral correction matrix. Then, it extracts low-pixel resolution images of 16 spectral channels and performs de-mosaic reconstruction using a weighted bilinear interpolation method to obtain full-pixel reconstructed images corresponding to the 16 spectral channels. Finally, based on the correspondence between the center wavelength of each spectral channel and time, the 16 full-pixel reconstructed images are converted into 16 time-series images.

[0009] This invention applies AOTF (Optical Aperture Transient Spectrum) to time-wavelength mapped ultra-high-speed imaging technology. By combining a supercontinuum light source with AOTF, it generates time-varying narrowband illumination, enabling spectral encoding of dynamic scenes. Combined with a hyperspectral camera for spectral resolution acquisition, spectral information can be efficiently converted into time-series images. The technical solution provided by this invention does not rely on an ultrafast laser light source, achieving ultra-high-speed imaging with a simple and compact structure, flexible time window, and high imaging fidelity, providing a reliable technical means for high-resolution real-time observation of complex transient phenomena.

[0010] The beneficial effects of this invention are: (1) This invention utilizes the synergistic effect of a broadband supercontinuum light source and an acousto-optic tunable filter (AOTF) to generate time-varying narrowband illumination light, thereby spectrally encoding the temporal evolution of a dynamic scene. Subsequently, a hyperspectral camera acquires the spectrally encoded dynamic scene in a single exposure and precisely extracts image information for each spectral channel based on pixel arrangement. Through this direct spectral-temporal mapping method, this scheme does not rely on reconstruction algorithms such as compressed sensing, thus avoiding the detail loss and artifact problems that may occur in traditional reconstruction processes, and achieving high-fidelity imaging of dynamic scenes. Under this design, this invention can simultaneously obtain an ultra-high-speed imaging frame rate of 1.6 million frames per second and a high spatial resolution of 724 lp / mm, achieving high-speed, high-resolution, and high-fidelity single-exposure ultra-high-speed imaging.

[0011] (2) The acousto-optic tunable filter (AOTF) used in this invention has the characteristic that the scanning time and filtering spectral range can be flexibly adjusted, enabling the invention to freely set the observation time window according to different experimental needs. At the same time, the AOTF does not require mechanical movement during frequency scanning, avoiding vibration and wear problems caused by mechanical parts, thus giving the system high operational stability and long-term reliability. In addition, the electronic drive method makes its response speed fast and repeatable, and can precisely control the spectral modulation process, ensuring that the imaging process can maintain consistency and high fidelity in various dynamic scenarios.

[0012] (3) The present invention has a simple and compact structure. The overall system mainly consists of a supercontinuum light source, an acousto-optic tunable filter (AOTF), and a hyperspectral camera. The number of components is small and the optical path layout is clear, which facilitates integration and debugging. Since the spectral modulation is achieved by electronic control, time-varying narrowband illumination can be generated without relying on an ultrafast laser light source, thereby significantly reducing the system construction and operation costs. Attached Figure Description

[0013] Figure 1 This is a structural diagram of the present invention; Figure 2 This is a schematic diagram illustrating the working principle of the acousto-optic tunable filter (AOTF) of the present invention; Figure 3 This is a flowchart of the forward imaging process of the present invention; Figure 4 This is a diagram showing the morphological evolution of water-in-oil droplets in a microfluidic channel as measured by this invention. In the figure: 100-Spectral modulation system; 101-Supercontinuum light source; 102-Short-pass filter; 103-Linear polarizer; 104-Acousto-optic tunable filter (AOTF); 105-First lens; 200-Spectral imaging system; 201-Sample to be tested; 202-Objective lens; 203-Second lens; 204-Hyperspectral camera; 300-Data processing system. Detailed Implementation

[0014] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0015] See Figure 1 The present invention includes: A spectral modulation system 100 consisting of a supercontinuum light source 101, a short-wave pass filter 102, a linear polarizer 103, an acousto-optic tunable filter AOTF 104, and a first lens 105; The supercontinuum light source 101, short-wave pass filter 102, linear polarizer 103, acousto-optic tunable filter AOTF 104, and first lens 105 of the spectral modulation system 100 are connected in sequence via optical paths. A spectral imaging system 200 consisting of a sample to be tested 201, an objective lens 202, a second lens 203, and a hyperspectral camera 204; The sample to be tested 201, objective lens 202, second lens 203 and hyperspectral camera 204 of the spectral imaging system 200 are connected in sequence via optical paths. A data processing system 300 composed of computers; The first lens 105 of the spectral modulation system 100 is connected to the optical path of the sample 201 of the spectral imaging system 200. The computer of the data processing system 300 is connected to the acousto-optic tunable filter AOTF 104 of the spectral modulation system 100 and the hyperspectral camera 204 of the spectral imaging system 200 via data cables.

[0016] The spectral modulation system 100 dynamically modulates broadband light through acousto-optic filtering to generate illumination light that varies with time, and projects the illumination light onto the surface of the sample 201 to be tested, thereby mapping the time evolution information of the sample 201 to the spectral domain.

[0017] The spectral imaging system 200 collects the illumination light that passes through the sample 201 and images it onto the hyperspectral camera 204 to obtain the spectral encoding spatiotemporal information of the sample 201.

[0018] The data processing system 300 processes the original mosaic image acquired by the hyperspectral camera 204. First, it suppresses crosstalk between spectral channels based on a preset spectral correction matrix. Then, it extracts low-pixel resolution images of 16 spectral channels and performs de-mosaic reconstruction using a weighted bilinear interpolation method to obtain full-pixel reconstructed images corresponding to the 16 spectral channels. Finally, based on the correspondence between the center wavelength of each spectral channel and time, the 16 full-pixel reconstructed images are converted into 16 time-series images.

[0019] This invention works as follows: See Figure 1 , Figure 2First, the supercontinuum light source 101 of the spectral modulation system 100 outputs broadband supercontinuum light, with a spectral coverage of 410nm-2400nm, a pulse width of approximately 6ps, a maximum power of 7W, and a repetition frequency adjustable between 10kHz and 80MHz. The broadband supercontinuum light is filtered by a short-pass filter 102 with a cutoff wavelength of 600nm, allowing only light with wavelengths less than 600nm to pass through, thus reducing the spectral range to 410nm-600nm. Next, it passes through a linear polarizer 103, and by adjusting its transmission direction, the transmitted broadband light is made linearly polarized, with the polarization direction along the vertical direction. Subsequently, the broadband light enters an acousto-optic tunable filter AOTF 104. In this system, the acousto-optic tunable filter AOTF 104 operates in swept-frequency mode for high-speed, time-sequential scanning filtering of the incident broadband light. By setting the scanning range and total scanning time of the radio frequency signal, for example, completing a linear scan of the 50nm band within 10μs, narrowband light with different center wavelengths can be output sequentially on the time axis, thereby forming a time-varying spectrum that evolves over time. In this invention, the scanning band is set to 450nm-600nm, and the total scanning time is flexibly adjusted according to different experimental scenarios. The time-varying illumination light emitted from the acousto-optic tunable filter AOTF 104 is focused by the first lens 105 and enters the spectral imaging system 200.

[0020] See Figure 1 The purpose of the short-pass filter 102 used in this invention is not only to limit the spectral range and filter out unwanted long-wavelength bands, but also to reduce the risk of damage to subsequent optical components caused by the high power output of the supercontinuum light source 101. Because the supercontinuum light source 101 has extremely high energy, failure to first reduce the power in the long-wavelength bands may cause the following problems: First, excessive light power may exceed the damage threshold of subsequent optical devices, affecting their normal operation or even causing permanent damage; second, excessively high light intensity will also increase the safety hazards to experimental personnel during system operation.

[0021] See Figure 1 The purpose of the linear polarizer 103 used in this invention is to ensure that the light incident on the acousto-optic tunable filter AOTF 104 is vertically polarized. Because the acousto-optic tunable filter AOTF 104 is selective for the polarization direction of the incident light, only light with a polarization direction consistent with the device design can achieve efficient modulation; light with other polarization directions will lead to a decrease in modulation efficiency.

[0022] See Figure 1 , Figure 2The acousto-optic tunable filter AOTF 104 used in this invention is a spectral modulation device based on the acousto-optic effect. Its working principle is to achieve wavelength-selective diffraction of incident light by adjusting the frequency of the radio frequency signal. This device has advantages such as high precision and fast response, and can support single or multiple radio frequency signals to achieve modulation of single or multiple bands. It can also achieve continuous band modulation through continuous radio frequency scanning. This invention utilizes its continuous frequency sweep function, setting the scanning band to 450nm-600nm to match the response range of the hyperspectral camera 204 used.

[0023] See Figure 1 The first lens 105 used in this invention is an achromatic cemented doublet lens, whose focal length can be selected or adjusted according to specific lighting requirements. Its function is to accurately focus the illumination light onto the surface of the sample 201 to achieve effective illumination of the target area.

[0024] See Figure 1 , Figure 3 The sample 201 to be tested in the spectral imaging system 200 can be a dynamically moving object or a stationary object. During system operation, if the sample 201 moves, it constitutes a dynamic scene. Under convergent time-varying illumination, the dynamic scene, evolving over time, is transformed into a spectral dimension through a time-spectral mapping relationship, forming a hyperspectral data cube. The original xy-space information remains unchanged, while the time dimension corresponds to wavelength changes. The mapped hyperspectral data is magnified and imaged onto the photosensitive surface of the hyperspectral camera 204 via the objective lens 202 and the second lens 203. The hyperspectral camera 204 acquires the complete spectrally encoded dynamic scene in a single exposure, obtaining a mosaic raw image with a size of 2048×1088 pixels. This image is transmitted to the computer of the data processing system 300 via a high-speed data interface.

[0025] See Figure 1 The objective lens 202 used in this invention is used to efficiently collect the light signal modulated by the sample 201 under test. The magnification can be selected according to experimental requirements and sample size. The second lens 203, as a tubular lens used in conjunction with the objective lens 202, is generally an achromatic cemented doublet lens with a focal length of 200mm.

[0026] See Figure 1The hyperspectral camera 204 used in this invention is based on a standard CMOS sensor and integrates a Fabry-Perot interferometer filter array at the wafer level. This design constructs a mosaic array composed of narrowband interferometer filters on top of the pixel structure, ensuring that each pixel or group of pixels transmits light signals only within a specific wavelength range. Specifically, the sensor covers the 450nm-600nm band, provides 16 independent spectral channels, and is periodically distributed in a 4×4 mosaic pattern throughout the pixel array, with a total pixel count of 2048×1088.

[0027] See Figure 1 After the experimental data acquisition is completed, the computer in the data processing system 300 needs to process the original mosaic image to accurately recover the time series information of the dynamic scene. First, due to crosstalk between spectral channels during data acquisition by the hyperspectral camera 204, this invention utilizes the spectral correction matrix provided by the manufacturer to perform pixel-level correction processing on the original observation signal. At each pixel location, the observed values ​​of the 16 channels are weighted and calculated with the correction matrix to recover the independent spectral response of each channel. Second, based on the spectral filter arrangement of the pixel array of the hyperspectral camera 204, the corrected image is separated into 16 independent spectral channels, and the corresponding 512×272 pixel low-resolution spectral image is extracted. This image is then de-mosaiced using a weighted bilinear interpolation algorithm to reconstruct a 2048×1088 pixel image, thus obtaining 16 full-resolution spectral channel images. Finally, the position of the center wavelength of each frame's spectral channel image across the entire spectral range is mapped onto the total time window to determine the occurrence time of each frame, thereby obtaining the time series image of the dynamic scene.

[0028] Example: This example is used to observe the dynamic morphological evolution of water-in-oil droplets in a microfluidic channel to verify the feasibility of the ultra-high-speed imaging device based on acousto-optic filtering modulation of the present invention.

[0029] See Figure 1 The device described in this embodiment includes a spectral modulation system 100, a spectral imaging system 200, and a data processing system 300.

[0030] The supercontinuum light source 101 of the spectral modulation system 100 outputs broadband supercontinuum light with a spectral range of 410nm-2400nm, a pulse width of approximately 6ps, a repetition frequency of 80MHz, and a power consumption of 4.5W. The short-pass filter 102 has a cutoff wavelength of 600nm, allowing only light with wavelengths less than 600nm to pass through. The linear polarizer 103 has its transmission direction perpendicular to the optical platform, ensuring that the transmitted light is vertically polarized linearly. The acousto-optic tunable filter AOTF 104 operates in continuous sweep mode, with the scanning band set to 450nm-600nm and the total scanning time set to 300μs. The first lens 105 is an achromatic cemented doublet with a focal length of 150mm and a diameter of 50.8mm.

[0031] The sample 201 of the spectral imaging system 200 is a microfluidic oil-in-water droplet device, mainly consisting of a PDMS microfluidic droplet chip, two syringe pumps, and mineral oil and water. One syringe pump injects mineral oil into both sides of the chip, while the other syringe pump introduces water from the middle. By adjusting the flow rate of the syringe pumps, the oil-to-water flow rate ratio is controlled at 5:2. At the cross-shaped junction of the chip, the oil phase applies shear and extrusion forces to the water phase, cutting the water flow into oil-in-water droplets. The droplets then enter the wide channel from the narrow channel, forming the dynamic scene observed in the experiment.

[0032] The objective lens 202 of the spectral imaging system 200 has a magnification of 20x. The second lens 203 is an achromatic cemented doublet with a focal length of 200mm and a diameter of 50.8mm, used in conjunction with the objective lens 202. The hyperspectral camera 204 covers the 450nm-600nm wavelength band, providing 16 independent spectral channels, which are periodically distributed in a 4×4 mosaic pattern throughout the pixel array, with a total pixel count of 2048×1088.

[0033] This embodiment works as follows: See Figure 1 , Figure 2A supercontinuum light source 101 outputs broadband supercontinuum light with a spectral range of 410 nm to 2400 nm. After passing through a short-pass filter 102 and a linear polarizer 103, it forms vertically polarized light with a spectral range of 410 nm to 600 nm. The linearly polarized light then enters an acousto-optic tunable filter (AOTF) 104, and after continuous frequency sweep filtering, outputs time-varying illumination light with a spectral range of 450 nm to 600 nm. Subsequently, the time-varying illumination light is focused onto the surface of the sample 201 under test by a first lens 105. Simultaneously, water-in-oil droplets are sheared within a microfluidic channel. The movement of the water-in-oil droplets in the microfluidic channel is magnified 20 times by the objective lens 202 and the second lens 203 and imaged onto a hyperspectral camera 204. The hyperspectral camera 204 acquires the spectrally encoded dynamic scene in a single exposure, obtaining a mosaic raw image with a size of 2048 × 1088 pixels.

[0034] See Figure 1 , Figure 4 After the experimental data was acquired and transmitted to the computer of the data processing system 300 via a high-speed data interface, the computer processed the acquired original mosaic image with a size of 2048×1088 pixels. First, the spectral correction matrix provided by the manufacturer was used to weight and correct the signals of each pixel to suppress crosstalk between spectral channels. Second, based on the filter arrangement of the 204-pixel array of the hyperspectral camera, 512×272 pixel low-resolution images corresponding to 16 spectral channels were extracted, and de-mosaic reconstruction was performed using a weighted bilinear interpolation method to generate a full-resolution spectral channel image of 2048×1088 pixels. Finally, based on the correspondence between the center wavelength of each spectral channel and the total scanning time, the 16 full-pixel reconstructed images were mapped to 16 time-series images, thus obtaining a dynamic sequence of the water-in-oil droplet motion with a total time span of 300 μs. Five frames were selected from the recovered 16 time-series images for display, such as... Figure 4 As shown in the figure, the morphological evolution of the water-in-oil droplet during its transport in the microfluidic channel can be clearly observed, fully verifying the feasibility and effectiveness of the device of the present invention in dynamic droplet imaging.

[0035] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-speed imaging device based on acousto-optic filtering modulation, characterized in that it include: A spectral modulation system (100) consisting of a supercontinuum light source (101), a short-pass filter (102), a linear polarizer (103), an acousto-optic tunable filter (AOTF) (104), and a first lens (105). The supercontinuum light source (101), short-wave pass filter (102), linear polarizer (103), acousto-optic tunable filter (AOTF) (104) and first lens (105) of the spectral modulation system (100) are connected in sequence via optical paths. A spectral imaging system (200) consisting of a sample to be tested (201), an objective lens (202), a second lens (203), and a hyperspectral camera (204). The sample to be tested (201), objective lens (202), second lens (203) and hyperspectral camera (204) of the spectral imaging system (200) are connected in sequence via optical paths; A data processing system (300) consisting of computers; The first lens (105) of the spectral modulation system (100) is optically connected to the sample (201) under test in the spectral imaging system (200); The computer of the data processing system (300) is connected to the acousto-optic tunable filter AOTF (104) of the spectral modulation system (100) and the hyperspectral camera (204) of the spectral imaging system (200) via data lines. Among them, the supercontinuum light source (101) outputs broadband supercontinuum light; the broadband supercontinuum light is filtered by a short-wave pass filter (102) to limit the spectral range, and then passes through a linear polarizer (103) and its transmission direction is adjusted so that the transmitted broadband light is linearly polarized light with the polarization direction along the vertical direction; subsequently, the vertically polarized broadband light enters the acousto-optic tunable filter AOTF (104), which operates in continuous frequency sweep mode. By setting the frequency sweep range and total sweep time of the radio frequency signal, the incident broadband light is subjected to high-speed time-sequential scanning filtering, thereby outputting narrowband light with different center wavelengths in sequence on the time axis, forming illumination light that changes rapidly with time.

2. The ultra-high-speed imaging device based on acousto-optic filtering modulation according to claim 1, characterized in that, The spectral modulation system (100) dynamically modulates broadband light through acousto-optic filtering to generate illumination light that changes over time, and projects the illumination light onto the surface of the sample to be tested (201), thereby mapping the time evolution information of the sample to be tested (201) to the spectral domain.

3. The ultra-high-speed imaging device based on acousto-optic filtering modulation according to claim 1, characterized in that, The spectral imaging system (200) collects the illumination light that passes through the sample to be tested (201) and images it onto the hyperspectral camera (204) to obtain the spectral encoding spatiotemporal information of the sample to be tested (201).

4. The ultra-high-speed imaging device based on acousto-optic filtering modulation according to claim 1, characterized in that, The data processing system (300) processes the original mosaic image acquired by the hyperspectral camera (204). First, it suppresses crosstalk between spectral channels based on a preset spectral correction matrix. Then, it extracts low-pixel resolution images of 16 spectral channels and performs de-mosaic reconstruction using a weighted bilinear interpolation method to obtain full-pixel reconstructed images corresponding to the 16 spectral channels. Finally, based on the correspondence between the center wavelength of each spectral channel and time, it converts the 16 full-pixel reconstructed images into 16 time-series images.

Images