A Mid-Wave Infrared High-Speed Hyperspectral Imaging Method

Abstract

The present invention discloses a mid-wave infrared high-speed hyperspectral imaging method, which is characterized in that the method includes: 1) acquiring two-dimensional spatial and spectral information of a sample to be measured; 2) spatially combining an infrared signal light and a pump light by using a dichroic mirror, and performing broadband frequency up-conversion to the visible / near-infrared band based on a chirped polarization crystal; 3) performing narrowband spectroscopy on the up-converted broadband signal by using an acousto-optic tunable filter to obtain narrowband images of different wavelengths; 4) calculating a scaling factor of the image according to the wavelength of the narrowband image and the focal length of the lens in the imaging system, and performing size correction on each narrowband image to obtain a final hyperspectral data cube and other steps. Compared with the prior art, the present invention has the characteristics of fast spectroscopy speed, high spectral resolution, high diffraction efficiency, etc., eliminates the dependence on mechanical scanning or temperature tuning, enables the imaging system to have the advantages of high spectral formation speed, high resolution, high sensitivity, etc., and greatly improves the image data acquisition rate.

Description

Technical Field

[0001] The present invention relates to the field of hyperspectral imaging technology, and in particular to a mid-wave infrared high-speed hyperspectral imaging method. Background Art

[0002] Hyperspectral imaging, as a means of obtaining multi-dimensional information, has the characteristic of integrating spectrum and image. It combines imaging technology and spectral technology, enabling the obtained spectral image data cube to contain both the image contour information and fine spectral information of the target object. In particular, the mid-infrared band has a relatively wide spectral range, covering multiple transmission windows of the Earth's atmosphere, containing the vibration-rotation energy level transition characteristic spectral lines of many important molecules, and having strong dust and fog penetration ability. Therefore, the development of high-performance mid-wave infrared hyperspectral imaging technology has always been an international research hotspot, and it has important applications in fields such as medical diagnosis, remote sensing exploration, and material detection.

[0003] However, mid-infrared hyperspectral imagers are restricted by traditional technologies and devices such as spectral splitting methods and infrared focal plane detection arrays, and it is difficult to balance the acquisition speed, sensitivity, and spectral resolution of the spectral image data cube. The acquisition of hyperspectral information by existing mid-infrared hyperspectral imagers usually relies on spectral splitting devices such as gratings and prisms, and cooperates with scanning devices to obtain the data cube. The required spatial distance is long, resulting in a large volume of the imaging device, and the imaging update speed is limited by the scanning device. The snapshot spectral imaging method can obtain a complete data cube within a single exposure period through multi-channel parallel acquisition combined with narrowband filter films. Its number of spectral channels is usually small, the spectral resolution is low, and it is usually greater than one hundred nanometers for the mid-infrared band. The acousto-optic tunable filter has the characteristics of fast spectral splitting speed, high spectral resolution, high diffraction efficiency, and small volume. It has developed rapidly in recent years. It balances the acquisition speed and spectral resolution of the data cube through fast spectral scanning. However, the acousto-optic tunable filter applied to the mid-infrared band has a low diffraction efficiency, and the image acquisition speed of mid-infrared imaging devices is much slower than the spectral splitting speed, and the data cube acquisition speed is still limited.

[0004] To obtain an effective response to mid-infrared low-energy photons, infrared detectors for mid-infrared hyperspectral imaging usually use semiconductor materials with a relatively narrow bandgap (such as mercury cadmium telluride and indium antimonide). At room temperature, infrared detectors generally have a large dark noise and usually require complex and expensive cooling devices, which greatly limits their sensitivity. In addition, the working frame rate of infrared detector arrays is usually dozens of fps. In contrast, existing silicon-based cameras have more advantages in performance. Their working frame rate can reach the order of Mfps, and they have extremely high sensitivity. However, limited by the bandgap energy of semiconductor materials, silicon-based cameras generally operate in the visible / near-infrared band and cannot directly respond to the mid-infrared band.

[0005] For this reason, the nonlinear frequency up-conversion technology provides a feasible solution. Usually, the infrared signal is converted to the visible / near-infrared band by the nonlinear sum-frequency process, so as to make full use of the excellent silicon-based devices to achieve high-sensitivity detection and imaging. However, limited by the strict phase matching conditions, the up-conversion imaging system faces problems that need to be solved urgently, such as narrow detection bandwidth and limited field of view angle. The existing technologies generally use methods such as tuning the phase matching parameters or using broadband illumination, which require post-stitching and registration of the image spatial and spectral information, resulting in long image acquisition time and complex post-processing. Therefore, it is still quite challenging to achieve large-field-of-view and high-speed hyperspectral imaging in the mid-infrared band based on the frequency up-conversion scheme. Summary of the Invention

[0006] The object of the present invention is to provide a mid-wave infrared high-speed hyperspectral imaging method and its imaging system for the deficiencies of the existing technology. The broadband frequency up-conversion method based on a nonlinear crystal is adopted to convert the large-field-of-view broadband mid-infrared signal into a near-infrared / visible light signal. The broadband up-conversion signal is quickly spectroscopically analyzed by an acousto-optic tunable filter, and combined with a high-speed silicon-based imaging device to realize real-time and high-speed acquisition and processing of mid-infrared spectral image data. By using a nonlinear crystal with a chirped polarization structure, the limitations of the traditional up-conversion imaging scheme, such as limited field of view angle and narrow matching wavelength, are broken through, and different wavelength components and different incident angles in the mid-infrared signal can be converted, effectively broadening the imaging field of view. Combining the frequency up-conversion method to convert the broadband infrared signal to the visible / near-infrared band, and using the characteristics of the acousto-optic tunable filter, such as fast spectroscopic analysis speed, high diffraction efficiency, and high spectral resolution, the imaging system has the advantages of fast response speed, high sensitivity, and high spectral resolution. A high-performance silicon-based camera is used for image acquisition, avoiding the deficiencies of the existing infrared detection and imaging devices in working frame rate and sensitivity, and eliminating the dependence on mechanical scanning or temperature tuning. The imaging system has the advantages of high spectral formation speed, high resolution, and high sensitivity, and has the characteristics of high acquisition speed and high sensitivity. The spectroscopic analysis speed is fast, the spectral resolution is high, and the diffraction efficiency is high, and it is easy to obtain high-speed and high-quality imaging results. The method is simple and has a wide application prospect.

[0007] The specific technical solution to achieve the object of the present invention is: a mid-wave infrared high-speed hyperspectral imaging method, which is characterized by adopting the broadband frequency up-conversion method of a nonlinear crystal to convert the large-field-of-view broadband mid-infrared signal into a near-infrared / visible light signal, quickly spectroscopically analyzing the broadband up-conversion signal by an acousto-optic tunable filter, and using a basic imaging device to realize the acquisition of real-time and high-speed mid-infrared spectral image data, specifically including the following steps:

[0008] Step 1: Use a broadband infrared signal light source to irradiate the sample to be measured to obtain the two-dimensional spatial and spectral information of the sample;

[0009] Step 2: Use a dichroic mirror to spatially combine the infrared signal light and the pump light, and convert the infrared signal to the visible / near-infrared band through broadband frequency up-conversion based on a chirped polarization crystal;

[0010] Step 3: Split the up-converted broadband signal through an acousto-optic tunable filter to obtain narrowband images of different wavelengths;

[0011] Step 4: Calculate the scaling factor of the image based on the wavelength of the narrowband image and the focal length of the lens in the imaging system, and correct the size of each narrowband image to obtain a hyperspectral data cube.

[0012] The imaging system includes: a broadband infrared light source, a pump light source, a broadband frequency up-conversion module, an acousto-optic tunable filter, an image acquisition module, and an image processing module; the broadband infrared light source irradiates the sample to be measured to obtain its two-dimensional image information and spectral information, and enters the broadband frequency up-conversion module after being combined with the pump light source to achieve an effective conversion from the infrared band to the visible light band. The broadband visible light signal obtained by frequency conversion is quickly split by the acousto-optic tunable filter. At the same time, each narrowband image is collected by the image acquisition module, and then the image size is corrected by the image processing module to obtain the final spectral image data cube.

[0013] The broadband infrared light source is broadband infrared light generated by supercontinuum generation, broadband infrared light generated by a thermal light source, or broadband infrared light generated by an optical parametric process.

[0014] The image acquisition module consists of a signal generator, a computer, and a silicon-based CMOS camera, and works in cooperation with the acousto-optic tunable filter. The specific process is as follows: The computer controls the signal generator to generate a triangular wave as a trigger signal to drive the acousto-optic tunable filter to work, and the speed of change of the acousto-optic drive frequency is determined by the slope of the triangular wave. At the same time, the signal generator generates a sawtooth wave as a trigger signal to drive the silicon-based CMOS camera to collect images at high speed, and the sawtooth wave and the triangular wave generated by the signal generator are synchronous signals.

[0015] The image processing module is implemented by a computer, aiming to calculate the scaling factor M of the up-converted image, and then realize image size correction. Specifically, for the scaling factor M corresponding to the k-th acquired image k is represented by the following formula (c):

[0016]

[0017] where f1 and f2 are the focal lengths of the plano-convex lens in front of the crystal and the plano-convex lens behind the crystal in the imaging 4f system, respectively, λ p is the pump wavelength, and λ k is the filtering wavelength of the acousto-optic tunable filter for the k-th acquired image.

[0018] The broadband frequency up-conversion module consists of a plano-convex lens, a chirped polarization nonlinear crystal and a resonant cavity, aiming to achieve broadband and efficient frequency up-conversion of mid-infrared signals. The plano-convex lens is used to focus and collimate mid-infrared signals; the chirped polarization nonlinear crystal has a broadband phase-matching window and can achieve the conversion of broadband infrared signals; the resonant cavity enhances the oscillation of the pump light power in the cavity, thereby improving the frequency up-conversion efficiency.

[0019] Computers are provided in both the above-mentioned image acquisition module and the image processing module. The computer in the image acquisition module is used to control the device to collect data in a linked manner, and the computer in the image processing module is used to calculate the image scaling factor and complete size correction. The computers in the two modules can share one to complete the above two functions.

[0020] The present invention has the following remarkable technical effects and improvements compared with the prior art:

[0021] 1) An acousto-optic tunable filter is used to split the broadband up-converted signal. Compared with the traditional grating- and prism-based splitting schemes, it eliminates the limitation of the mechanical scanning device on the imaging update rate. Its drive frequency switching time is as low as the ns level, which helps to greatly improve the image data acquisition rate and achieve high-speed mid-infrared hyperspectral imaging.

[0022] 2) An acousto-optic tunable filter is used to split the broadband up-converted signal. Compared with the traditional snapshot multi-channel parallel imaging scheme, the spectral resolution is increased by one order of magnitude, the number of channels is greatly increased, and the spectral details of the data cube are richer, which helps to achieve high-resolution mid-infrared hyperspectral imaging.

[0023] 3) The frequency up-conversion technology is used to collect images with a high-performance silicon-based camera, avoiding the defects of large intrinsic dark noise and the need for cryogenic cooling operation in existing infrared detection and imaging devices. The sensitivity is greatly improved and can even reach the single-photon level, which helps to achieve high-sensitivity mid-infrared hyperspectral imaging.

[0024] 4) A nonlinear crystal with a chirped polarization structure is used to achieve the effective conversion of broadband mid-infrared signals. At the same time, infrared signals incident at different angles can be effectively matched to different inversion periods, thereby broadening the acceptance angle of the crystal for incident light and achieving large-field-of-view mid-infrared hyperspectral imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a schematic diagram of the present invention;

[0026] Figure 2 is a schematic diagram of a mid-wave infrared high-speed hyperspectral imaging system;

[0027] Figure 3 is a schematic diagram of an embodiment. Detailed implementation mode

[0028] Referring to Figure 1 , the present invention realizes broadband frequency up-conversion based on a chirped polarization structure nonlinear crystal, converts a large field-of-view broadband mid-infrared signal into a near-infrared / visible light signal, then quickly disperses the broadband up-converted signal through a narrowband filter, and combines a high-speed silicon-based imaging device to realize real-time high-speed mid-infrared spectral image data acquisition and processing.

[0029] The realization of broadband frequency up-conversion by the nonlinear crystal based on the chirped polarization structure has the following functions:

[0030] 1) Different wavelength components in the signal light require different polarization periods and temperatures of the crystal to satisfy the quasi-phase matching condition, and the chirped polarization structure can convert each wavelength component in the mid-infrared signal;

[0031] 2) Different incident angles in the signal light require different polarization periods and temperatures of the crystal to satisfy the quasi-phase matching condition. The chirped polarization structure can effectively convert different incident angles of each narrowband component in the broadband mid-infrared signal light, and realize up-conversion imaging with a large field of view;

[0032] 3) Compared with a nonlinear crystal with a single polarization period to reduce the crystal length to obtain a larger phase matching bandwidth, the chirped polarization structure has a higher conversion efficiency under the same phase matching reception bandwidth, taking into account the sensitivity of the up-conversion imaging system.

[0033] The chirped polarization crystal of the present invention broadens the phase matching bandwidth in the up-conversion process, realizes up-conversion in a wide spectral range and wide-field imaging, and breaks through the limitations of limited field of view angle and narrow matching wavelength in traditional up-conversion imaging schemes. In addition, based on the chirped polarization crystal, a large field-of-view image is converted at one time, eliminating processes such as mechanical scanning and temperature tuning in traditional schemes, which helps to greatly improve the image data acquisition rate.

[0034] The narrowband filter quickly disperses the broadband up-converted signal. The existing visible / near-infrared acousto-optic tunable filter has high spectral resolution (~1nm), high diffraction efficiency (~80%), and fast dispersion speed (~8ns). Combining with the frequency up-conversion technology, it can make full use of the high-performance silicon-based cameras in the existing visible / near-infrared bands, and cooperate with the acousto-optic tunable filter to quickly disperse the light to realize high-speed hyperspectral image acquisition, and at the same time has the advantages of high acquisition speed and high sensitivity.

[0035] The present invention adopts a broadband frequency up-conversion method based on a nonlinear crystal. Different wavelength components in the broadband signal will introduce a different scaling factor, and the image ratio after dispersion needs to be corrected. Specifically, the image scaling factor of the up-converted image compared with the original image is expressed by the following formula (a):

[0036] M = λ u f2 / λ s f1(a).

[0037] Wherein, λ u is the center wavelength of the up-conversion light; λ s is the wavelength of the corresponding incident infrared signal light; f2 is the focal length of the post-focusing lens of the crystal; f1 is the focal length of the pre-focusing lens of the crystal.

[0038] Since in frequency up-conversion, the signal light wavelength, pump light wavelength, and up-conversion light wavelength satisfy the following relationship in equation (b):

[0039]

[0040] The broadband up-conversion signal is split by an acousto-optic tunable filter to obtain narrowband signals with center wavelengths of λ1, λ2...λ k Combined with equation (b), the image scaling factor M k can be further converted into the following representation in equation (c):

[0041]

[0042] And by dividing the image size of each narrowband signal by the scaling factor M k , high-spectral image ratio correction can be achieved.

[0043] Refer to Figure 2 , the imaging system of the present invention includes: a broadband infrared light source, a pump light source, a broadband frequency up-conversion module, an acousto-optic tunable filter, an image acquisition and processing module. The broadband infrared light source irradiates the sample to be measured to obtain its two-dimensional image information and spectral information, and enters the broadband frequency up-conversion module through beam combination with the pump light source to achieve effective conversion from the infrared band to the visible light band. The wide-wavelength visible light signal obtained by frequency conversion is quickly split by the acousto-optic tunable filter. At the same time, each narrowband image is collected by the image acquisition module, and then the image size is corrected by the image processing module to obtain the final spectral image data cube. The mid-wave infrared high-speed high-spectral imaging specifically includes the following steps:

[0044] Step 1: Make the broadband infrared signal light source irradiate and pass through the sample to be measured to obtain the two-dimensional spatial and spectral information of the sample;

[0045] Step 2: Use a dichroic mirror to spatially combine the infrared signal light and the pump light, and convert the infrared signal to the visible / near-infrared band through a broadband frequency up-conversion process based on a chirped polarization crystal;

[0046] Step 3: Split the up-converted broadband signal by the acousto-optic tunable filter to obtain narrowband images of different wavelengths;

[0047] : Step 4: Calculate the scaling factor of the image based on the wavelength of the narrowband image and the focal length of the lens in the imaging system, and correct the size of each narrowband image to obtain the final hyperspectral data cube.

[0048] The broadband infrared light source is preferably, but not limited to, the following light source preparation methods: supercontinuum generation, thermal light source, or optical parametric process.

[0049] The present invention will be further described in detail below through specific embodiments.

[0050] Embodiment 1

[0051] Refer to Figure 3 , the mid-wave infrared high-speed hyperspectral imaging system includes: broadband infrared light source 1, first plano-convex lens 2, sample to be measured 3, second plano-convex lens 4, dichroic mirror 5, pump light source 6, first concave mirror 7, chirped poled lithium niobate crystal 8, second concave mirror 9, third plano-convex lens 10, filter 11, acousto-optic tunable filter 12, silicon-based camera 13, signal generator 14, computer image processing 15.

[0052] The broadband infrared light source 1 is an infrared wide-spectrum tungsten light source, and its output wavelength covers 450 - 5500 nm. Using this wide-spectrum light source as the detection light source, spectral imaging within the wavelength range of 2.5 - 5 μm of the sample can be achieved.

[0053] The first plano-convex lens 2 is a CaF2 lens, and its purpose is to collimate the mid-infrared signal light output by the wide-spectrum tungsten light source. The focal length of this plano-convex lens is 50 mm, and the lens diameter is 50.8 mm.

[0054] The sample to be measured 3 is the detection target, including but not limited to: biological tissue, tumor cells, or chemical materials. The absorption of light of different wavelengths by this detection target is different. The experimental system can obtain the absorption rate of each pixel of the detection target for each wavelength by measuring the imaging results with and without the sample.

[0055] The second plano-convex lens 4 is a CaF2 lens, and its purpose is to focus the infrared wide-spectrum light source passing through the target to be measured into the nonlinear frequency up-conversion medium to achieve efficient frequency conversion. The focal length of this plano-convex lens is 50 mm, and the lens diameter is 50.8 mm.

[0056] The dichroic mirror 5 spatially combines the broadband infrared light source (wide-spectrum infrared light source) 1 passing through the target to be measured and the pump light source 6, facilitating the subsequent generation of broadband frequency up-conversion signals. This dichroic mirror is a 2-μm long-wave pass dichroic mirror, which filters out the components below 2.5 μm in the signal light while spatially combining to prevent the wavelength components in this section from being confused with the up-converted signal components.

[0057] The pump light source 6 is a high-power 1 μm continuous laser with an output power of up to 10 W. It is used as the pump light of the broadband frequency up-conversion part to achieve effective conversion of the broadband infrared light source.

[0058] The first concave mirror 7 is a CaF2 lens, which aims to enhance the oscillation of the 1μm pump light source 6 in the cavity, thereby improving the broadband frequency conversion efficiency. It has high transmittance for 2.5-5μm and 0.7-0.9μm, and a reflectivity of 97% for 1μm.

[0059] The chirped polarized lithium niobate crystal 8 is used as a nonlinear frequency up-conversion medium to complete the effective conversion of the wide-spectrum infrared light source 1. Its polarization period covers 16 to 24 μm, and the chirped polarization step length is 0.01 mm. The crystal size is 5 mm (length) × 3 mm (width) × 1 mm (thickness). Under 1 μm high-power laser pumping, the chirped crystal polarization period used in this embodiment can realize the conversion of 2.5 to 5 μm infrared signals within the range of 0 to 60° of the incident full angle.

[0060] The second concave mirror 9 is a CaF2 lens, which is used to enhance the oscillation of the 1 μm pump light source 6 in the cavity, thereby improving the broadband frequency conversion efficiency. It has high transmittance for 2.5-5 μm and 0.7-0.9 μm, and a reflectivity of 97% for 1 μm.

[0061] The third plano-convex lens 10 is a CaF2 lens, and its purpose is to spatially collimate the up-conversion signal light. The focal length of the plano-convex lens is 50 mm, and the lens diameter is 50.8 mm.

[0062] The filter 11 is a bandpass filter with a transmission wavelength of 700-900 nm. The filter is used for up-conversion signal filtering to filter out high-power 1 μm pump light, pump light up-conversion fluorescence, ambient stray light, and the like.

[0063] The acousto-optic tunable filter 12 has an operating wavelength of 550-1000 nm, a spectral resolution of 2 nm, a frequency switching time of ≥8 ns, and a diffraction efficiency of >80% within the operating wavelength range.

[0064] The silicon-based camera 13 is a silicon-based COMS camera, which is used to collect narrow-band images after being split by an acousto-optic tunable filter at high speed, and its working frame rate is 10 kfps.

[0065] The purpose of the signal generator 14 is to generate a trigger signal to drive the AOT filter 12 and the silicon-based camera 13 to work.

[0066] The purpose of the computer image processing 15 is to calculate different image scaling factors of the up-converted image due to different incident wavelengths, and perform size correction on each narrow-band image after up-conversion.

[0067] The specific implementation process of mid-wave infrared high-speed hyperspectral imaging is as follows:

[0068] 1) A non-linear frequency up-conversion process is carried out between the broadband infrared light source (infrared broadband tungsten light source) 1 after obtaining the information of the sample to be measured 3 and the pump light source 6, generating broadband visible / near-infrared signal light. Specifically, the infrared broadband tungsten light source 1 is collimated by the first plano-convex lens 2 and then passes through the sample to be measured 3. After obtaining the spatial and spectral information of the sample, it is focused by the second plano-convex lens 4 and then passes through the dichroic mirror 5 to be spatially combined with the high-power 1-μm pump light source 6 and enters the chirped poled lithium niobate crystal 8. To make the broadband frequency up-conversion process have a higher conversion efficiency, the 1-μm pump light source 6 is made to oscillate back and forth through the concave cavity mirrors 7 and 9 to form a resonant cavity. The visible / near-infrared signal light generated by the non-linear frequency up-conversion process is spatially collimated by the third plano-convex lens 10. Among them, the temperature of the chirped poled lithium niobate crystal 8 is set at 30 °C, and the used plano-convex lenses 2, 4, and 10 are all CaF2 lenses, with a transmittance of >95% for the mid-infrared band. In the 4f imaging system composed of the sample to be measured 3, the second plano-convex lens 4, the third plano-convex lens 10, and the silicon-based CMOS camera 13, the pump beam is equivalent to a low-pass filter in the Fourier plane, filtering out some high-frequency components. Therefore, a larger pump beam diameter can obtain a higher imaging spatial resolution.

[0069] 2) Narrow-band fast filtering is performed on the preliminarily filtered broadband up-converted signal, and the filtered narrow-band image is recorded by the silicon-based COMS camera 13. Specifically, the up-converted signal is filtered by the band-pass filter 11 to filter out the 1-μm pump light, the up-converted fluorescence of the pump light, the ambient stray light, etc. The preliminarily filtered broadband up-converted signal enters the acousto-optic tunable filter 12 for narrow-band fast spectral splitting. The up-conversion center wavelength range of the 2.5-5-μm signal light and the 1-μm pump light is 715-833 nm. The acousto-optic tunable filter 12 has 120 bands in this wavelength range, and the spectral resolution is 1 nm. The computer controls the signal generator 14 to generate a triangular wave as a trigger signal to drive the acousto-optic tunable filter 12 to work, and the acousto-optic driving frequency change speed is determined by the slope of the triangular wave. At the same time, the signal generator 14 generates a sawtooth wave as a trigger signal to drive the silicon-based CMOS camera 13 to collect images at high speed, and the sawtooth wave and the triangular wave generated by the signal generator 14 are synchronous signals.

[0070] 3) By calculating the monochromatic image scaling factor M corresponding to each wavelength k, size correction is performed on all images to obtain the final hyperspectral cube data. Specifically, if the AOTF spectral scanning direction is from 715 nm to 833 nm, the scaling factor M corresponding to the k-th image k is expressed by the following formula (c):

[0071]

[0072] wherein, since the focal lengths f1 and f2 of the plano-convex lenses used in the 4f system are both 50 mm, and λ p is the fixed pump wavelength of 1 μm, the central wavelength λ corresponding to the k-th image k is (714 + k) nm, and the scaling factor M corresponding to the k-th image can be obtained k is expressed by the following formula (c-1):

[0073]

[0074] Each narrowband image is divided by the scaling factor M through the computer image processing module 15 k to correct the imaging size distortion, obtain the true image size, and further obtain the final hyperspectral data cube.

[0075] In this embodiment, by combining the broadband frequency up-conversion process and the rapid spectroscopy of the acousto-optic tunable filter 12, high-speed mid-infrared hyperspectral imaging is realized, and the effective wavelength conversion range is 2.5 - 5 μm. The spectral resolution of the acousto-optic tunable filter 12 is 1 nm, the number of hyperspectral acquisition channels is 120, and the actual resolution corresponding to the mid-infrared signal is 12 - 36 nm. In this embodiment, the acquisition speed of the hyperspectral cube data is limited by the silicon-based CMOS camera 13. The working frame rate of the camera is 10 kfps, and the time taken to acquire a single frame is 100 μs. Then, it only takes 12 ms to acquire the hyperspectral data cube. If a scientific research-level high-speed silicon-based imaging system with a working frame rate of 1 Mfps is further used in combination with the present invention, the acquisition time of the mid-infrared spectral image data cube is expected to be as low as 120 μs. Thanks to the chirped polarization structure lithium niobate crystal, the full acceptance angle of the crystal for incident light is 60°, and the imaging field of view diameter reaches 5 cm. By further optimizing the crystal polarization period and lens size, a larger imaging field of view will be obtained. The mid-wave infrared hyperspectral imaging system implemented based on the method of the present invention has the advantages of high acquisition speed, high spectral resolution, high sensitivity, and large field of view.

[0076] The above is only a specific embodiment of the present invention and is not intended to limit the present invention. The above embodiments have described the present invention in detail. The content not described in detail belongs to the prior art well-known to those skilled in the art. Those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the embodiments or perform equivalent replacements on some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the present invention. Any variations and improvements made without departing from the concept of the present invention fall within the scope of the claims of the present invention.

Claims

1. A mid-wave infrared high-speed hyperspectral imaging method, characterized in that A broadband frequency up-conversion method using a nonlinear crystal is adopted to convert a large field-of-view broadband mid-infrared signal into a near-infrared / visible light signal. The broadband up-converted signal is rapidly spectro-dispersed by an acousto-optic tunable filter, and a silicon-based imaging device is used to realize the acquisition of real-time high-speed mid-wave infrared spectral image data, which specifically includes the following steps: Step 1: Irradiate the sample to be measured with a broadband infrared signal light source to obtain the two-dimensional spatial and spectral information of the sample; Step 2: Use a dichroic mirror to spatially combine the infrared signal light and the pump light, and through broadband frequency up-conversion based on a chirped polarization crystal, convert the infrared signal to the visible / near-infrared band; Step 3: Spectro-disperse the up-converted broadband signal through an acousto-optic tunable filter to obtain narrowband images of different wavelengths; Step 4: Calculate the scaling factor of the image according to the wavelength of the narrowband image and the focal length of the lens in the imaging system, and correct the size of each narrowband image to obtain a hyperspectral data cube.

2. The mid-wave infrared high-speed hyperspectral imaging method according to claim 1, characterized in that The imaging system includes: a broadband infrared light source, a pump light source, a broadband frequency up-conversion module, an acousto-optic tunable filter, an image acquisition module, and an image processing module. The broadband infrared light source irradiates the sample to be measured to obtain its two-dimensional image information and spectral information, and enters the broadband frequency up-conversion module after being combined with the pump light source to realize the effective conversion from the infrared band to the visible light band. The wide-wavelength visible light signal obtained by frequency conversion is rapidly spectro-dispersed by the acousto-optic tunable filter; the image acquisition module acquires each narrowband image, and then the image processing module corrects the image size to obtain the final spectral image data cube.

3. The mid-wave infrared high-speed hyperspectral imaging method according to claim 2, characterized in that The broadband infrared light source is broadband infrared light generated by supercontinuum generation, broadband infrared light generated by a thermal light source, or broadband infrared light generated by an optical parametric process.

4. The mid-wave infrared high-speed hyperspectral imaging method according to claim 2, characterized in that The image acquisition module consists of a signal generator, a computer, and a silicon-based CMOS camera, and works in cooperation with the acousto-optic tunable filter. The specific process is as follows: The computer controls the signal generator to generate a triangular wave and a sawtooth wave synchronized with the triangular wave. The triangular wave is a trigger signal to drive the acousto-optic tunable filter to work, and the speed of change of the acousto-optic driving frequency is determined by the slope of the triangular wave; the sawtooth wave is a trigger signal to drive the silicon-based CMOS camera to acquire images at high speed.

5. The mid-wave infrared high-speed hyperspectral imaging method according to claim 2, characterized in that The image processing module obtains the scaling factor M of the up-converted image through a computer to achieve image size correction. For the scaling factor M corresponding to the k-th acquired image k is expressed by the following formula (c): where f1 and f2 are the focal lengths of the plano-convex lenses in front of and behind the crystal in the imaging 4f system, respectively; λ p is the pump wavelength; λ k is the filtering wavelength of the acousto-optic tunable filter for the k-th acquired image.

6. The method for mid-wave infrared high-speed hyperspectral imaging according to claim 2, wherein The broadband frequency up-conversion module consists of a plano-convex lens, a chirped polarization nonlinear crystal, and a resonant cavity, and realizes broadband and efficient frequency up-conversion of mid-infrared signals.

Images