Air cavity type optical filter, mid-infrared spectrum video computation camera device and camera method

Abstract

The application discloses a gas cavity type filter, a mid-infrared spectrum video computing camera device and a camera shooting method. The camera device comprises: an array type multispectral signal acquisition unit, which is used for acquiring original multispectral signals; each spectral signal acquisition unit is sequentially arranged along an optical axis and is composed of a gas cavity type filter, an infrared lens and an infrared detector; the gas cavity type filter is composed of an optical window, a metal shell, a filling gas, a gas filling interface and a sealing element, and the spectral transmittance of the filter can be adjusted by changing the parameters of the filling gas; a spatial registration unit, which is used for spatial registration processing of the original multispectral signals to obtain registered multispectral signals; and a spectral reconstruction unit, which is used for spectral reconstruction processing of the registered multispectral signals to obtain reconstructed hyperspectral signals. The application realizes the collaborative acquisition of high-resolution information in time, space and spectral dimensions, and guarantees the acquisition and reconstruction of high signal-to-noise ratio spectral signals.

Description

Technical Field

[0001] This application relates to the field of spectral imaging technology, and in particular to a gas cavity filter, a mid-infrared spectral video computing camera device and imaging method. Background Technology

[0002] Spectroscopic video imaging technology can acquire temporal, spatial, and spectral information of a scene, and has wide applications in fields such as chemical safety and agricultural remote sensing. In particular, the mid-infrared band (3μm-14μm) covers the characteristic absorption peaks of most gas molecules and is the core band for gas leak detection and material composition analysis.

[0003] Currently, mainstream mid-infrared hyperspectral imaging technologies mainly employ grating-based, prism-based, or solid-state filter-based beam splitting methods. Among these, grating and prism-based systems are complex in structure, bulky, and slow in imaging speed, making real-time video-level imaging impossible. While solid-state filter-based multispectral imaging systems offer faster imaging speeds, they suffer from the following drawbacks:

[0004] 1) Traditional multispectral imaging systems have a limited number of spectral channels and usually use narrowband filters, resulting in low spectral resolution, which cannot meet the needs of high-precision material identification.

[0005] 2) Solid-state filters are difficult to customize, have complex processes, and are costly to produce, especially wide-band filters with complex spectral transmittance curves, where the cost of a single filter can reach tens of thousands of yuan.

[0006] 3) The spectral characteristics of solid-state filters cannot be changed once they are determined, resulting in poor customization flexibility and difficulty in quickly adjusting the spectral channels for different application scenarios;

[0007] Therefore, how to achieve video-level hyperspectral imaging in the mid-infrared band while significantly reducing system costs and improving customization flexibility is a technical problem that this application urgently needs to solve. Summary of the Invention

[0008] The purpose of this application is to propose a cavity filter, a mid-infrared spectral video computational camera device and imaging method. It uses an array-type multispectral signal acquisition unit based on a cavity filter to acquire the original multispectral signal, and reconstructs the hyperspectral signal based on the computational imaging principle through spatial registration and broadband decoupling algorithms. It is suitable for video-level hyperspectral imaging applications in the mid-infrared band and has the advantages of simple structure, convenient customization and high signal-to-noise ratio.

[0009] The present invention provides a gas cavity filter, comprising: a housing, wherein optical window mounting positions are respectively provided at both ends of the housing; two optical windows are respectively sealed and mounted at the optical window mounting positions at both ends of the housing, forming a sealed cavity with the housing; and an inflation port is provided on the housing and communicates with the sealed cavity for filling the sealed cavity with filling gas.

[0010] In one possible implementation, the housing is a metal shell; the optical window is an optical glass window; a seal is provided at the connection between the optical glass window and the metal shell; and the inflation port is located on the side wall of the metal shell.

[0011] The metal casing has good mechanical strength and airtightness, and can withstand a certain range of air pressure changes; the optical glass window has good transmittance in the mid-infrared band; the seals ensure the long-term sealing of the air chamber; and the side wall inflation interface facilitates connection to external air sources and gas replacement.

[0012] This invention also provides a mid-infrared spectral video computing camera device, comprising: an array-type multispectral signal acquisition unit for acquiring raw multispectral signals, the array-type multispectral signal acquisition unit comprising N spectral signal acquisition units, where N is a positive integer greater than 1; each spectral signal acquisition unit is composed of a cavity filter, an infrared lens, and an infrared detector arranged sequentially along the optical axis; wherein, the cavity filter comprises a housing, optical windows respectively sealed at both ends of the housing, and an inflation port disposed on the housing and communicating with the sealed cavity, the optical windows and the housing forming the sealed cavity, and the inflation port for filling the sealed cavity with filling gas; the cavity filter is used to provide spectral transmittance in the target wavelength range, and the spectral transmittance of the cavity filter can be adjusted by changing the parameters of the filling gas; a spatial registration unit for performing spatial registration processing on the raw multispectral signals to obtain registered multispectral signals; and a spectral reconstruction unit for performing spectral reconstruction processing on the registered multispectral signals to obtain reconstructed hyperspectral signals.

[0013] As shown above, this device can acquire and reconstruct hyperspectral video signals in the mid-infrared band.

[0014] In one possible implementation, the housing is a metal shell; the optical window is an optical glass window; a seal is provided at the connection between the optical window and the metal shell; and the inflation port is located on the side wall of the metal shell.

[0015] Another aspect of the present invention provides a mid-infrared spectral video calculation camera method, applicable to the aforementioned mid-infrared spectral video calculation camera device, comprising: S10, acquiring N original multispectral signals through N spectral signal acquisition units of an array-type multispectral signal acquisition unit; S20, performing spatial registration processing on the original multispectral signals to obtain registered multispectral signals; S30, performing spectral reconstruction processing on the registered multispectral signals to obtain reconstructed hyperspectral signals.

[0016] In one possible implementation, step S00 is further included to pre-select the air-cavity filters for the N spectral signal acquisition units. One filling gas, specifically including:

[0017] S01: Select from the gas database according to preset rules A gas sample was used to construct an array-type multispectral signal acquisition unit;

[0018] S02: Acquire the raw multispectral signal through an array-type multispectral signal acquisition unit:

[0019] S03: Perform spatial registration processing on the original multispectral signal to obtain the registered multispectral signal;

[0020] S04: Perform spectral decoupling and reconstruction on the registered multispectral signal to obtain the reconstructed hyperspectral signal;

[0021] S05: Calculate the peak signal-to-noise ratio of the original hyperspectral signal and the reconstructed hyperspectral signal;

[0022] S06: Repeat steps S01 to S05. Second-rate, It is a positive integer. Find the peak signal-to-noise ratio corresponding to the maximum value. Each gas sample was used as... The filling gas in a gas cavity filter.

[0023] In one possible implementation, the spectral transmittance of the air cavity filter is calculated using the following formula:

[0024]

[0025] in, The spectral transmittance of the air-cavity filter is [value missing]. The spectral transmittance of the optical window. The spectral transmittance of the filling gas. For wavelength, .

[0026] As described above, the spectral transmittance of a cavity filter is mainly determined by the combined spectral transmittance of the optical window and the filling gas. Typically, once the optical window is selected and the overall structure of the cavity filter is fabricated, the spectral transmittance is primarily determined by the spectral transmittance of the filling gas. Therefore, the spectral transmittance of the filter can be adjusted by changing the parameters of the filling gas (including its type, concentration, and pressure).

[0027] In one possible implementation, the original multispectral signal is calculated using the following formula:

[0028]

[0029] in, The original multispectral signal, The original hyperspectral signal, This refers to the spectral transmittance of the infrared lens. This represents the spectral responsivity of the infrared detector.

[0030] In one possible implementation, the registered multispectral signal is calculated using the following formula:

[0031]

[0032] in, For the registered multispectral signal, This represents the spatial registration transformation operator.

[0033] In one possible implementation, the reconstructed hyperspectral signal is calculated using the following formula:

[0034]

[0035] Where H is the reconstructed hyperspectral signal, This represents the spectral reconstruction operator, where S is the number of target hyperspectral channels.

[0036] In summary, this invention employs an array-type multispectral signal acquisition unit based on a cavity filter to acquire the original multispectral signal and efficiently reconstructs the hyperspectral signal through an algorithm, offering the following advantages:

[0037] (1) Synergistic acquisition of temporal, spatial, and spectral information: Unlike traditional narrowband filters, the gas cavity filter provides broadband modulation and, combined with a broadband decoupling algorithm, achieves high-precision reconstruction of multispectral signals into hyperspectral signals, effectively improving the number of spectral channels and spectral resolution. Furthermore, unlike scanning spectral imaging systems, this invention employs an array architecture, enabling simultaneous acquisition of multiple spectral signals to achieve real-time video-level hyperspectral imaging, suitable for complex application scenarios such as dynamic gas leak detection in chemical industries.

[0038] (2) Reduced cost and flexible customization: The present invention uses a gas cavity filter to replace the traditional solid-state filter. The structure is simple and stable, which greatly reduces the design and processing difficulty of complex broadband filters. At the same time, the spectral transmission characteristics can be adjusted by changing the type and concentration of the gas filled in the cavity. There is no need to reprocess optical components, and it can be quickly adapted to different application scenarios.

[0039] (3) High signal-to-noise ratio signal reconstruction: Based on the array-type multispectral signal acquisition unit, this invention also provides an automatic selection method for gas cavity filters based on PSNR optimization, which can automatically select the optimal combination of gas filters to ensure the reconstruction accuracy of hyperspectral signals. Attached Figure Description

[0040] Figure 1 The diagram shown is a schematic of the structure of a cavity filter.

[0041] Figure 2 The diagram shown is a schematic representation of the overall structure of the infrared spectral video calculation camera device of the present invention.

[0042] Figure 3 The diagram shows the structure of an array-type multispectral signal acquisition unit.

[0043] Figure 4 The flowchart shown is a process for calculating infrared spectral video in this invention.

[0044] Figure 5 The diagram shows a flowchart of the gas selection method in the gas cavity filter of the present invention.

[0045] Figure 6 The figure shows a schematic diagram of the spectral transmittance curve of the gas filling the gas in the gas cavity filter.

[0046] The following are the labels in the diagram: 1-Housing, 3-Optical window, 4-Seal, 5-Inflation port, 6-Sealed cavity, 100-Array-type multispectral signal acquisition unit, 110-Multispectral signal acquisition unit, 111-Air cavity filter, 112-Infrared lens, 113-Infrared detector, 200-Spatial registration unit, 300-Spectral reconstruction unit. Detailed Implementation

[0047] Example 1

[0048] like Figure 1 As shown, this embodiment provides a cavity filter 111, including a housing 1, two optical windows 3, an inflation interface 5, and a sealing element 4.

[0049] Optical windows 3 consist of two pieces, each sealed and mounted at one end of the metal casing. They are made of germanium and provide high spectral transmittance in the long-wave infrared range (typically 7μm~12μm). Typical values ​​are usually above 95%;

[0050] The housing 1 is a metal shell made of aluminum, with optical window mounting positions at both ends;

[0051] The filling gas is sealed within the sealed cavity 6 formed by the housing 1 and the two optical windows 3;

[0052] The inflation port 5, containing a valve core, is located on the side wall of the metal casing 1 and communicates with the sealed cavity 6. It is used to fill the sealed cavity 6 with filling gas. The inflation port 5 can be made of brass and uses a standard gas nozzle connector for easy connection to an external gas source or vacuum pump. Through the inflation port 5, a specific type of filling gas 7 can be filled into the sealed cavity 6, or the existing gas can be extracted and replaced with a new gas.

[0053] A seal 4 is located at the connection between the optical window 3 and the housing 1 to ensure the airtightness of the sealed cavity 6. The seal 4 can be a rubber sealing ring, a polytetrafluoroethylene gasket, or a metal sealing gasket. For applications requiring frequent gas replacement, a ring seal structure that can be repeatedly disassembled and reassembled is preferred; for applications requiring long-term stable operation, welding seals or adhesive seals can be used.

[0054] Example 2

[0055] like Figure 2 As shown, this embodiment provides a mid-infrared spectral video computing camera device, including: an array-type multispectral signal acquisition unit 100, a spatial registration unit 200, and a spectral reconstruction unit 300. The array-type multispectral signal acquisition unit 100 includes N spectral signal acquisition units 110.

[0056] The array-type multispectral signal acquisition unit 100 is used to acquire raw multispectral signals.

[0057] Figure 3 The structural diagram of the array-type multispectral signal acquisition unit is as follows: Figure 3 As shown, the array-type multispectral signal acquisition unit 100 includes N spectral signal acquisition units 110, where N is a positive integer greater than 1. Each spectral signal acquisition unit 110 is composed of a cavity filter 111, an infrared lens 112, and an infrared detector 113 arranged sequentially along the optical axis.

[0058] The mid-infrared radiation emitted by the target scene is first filtered by the gas cavity filter 111, then focused by the infrared lens 112, and finally received by the infrared detector 113 and converted into an electrical signal. The gas cavity filters 111 of the N channels are filled with different types or concentrations of filling gas, so that each channel has different spectral transmission characteristics, thus forming N multispectral measurement channels with different spectral responses.

[0059] The cavity filter 111 is used to provide spectral transmittance in the target wavelength range, and the spectral transmittance of the cavity filter can be adjusted by changing the parameters of the filling gas.

[0060] Specifically, in this embodiment, the N cavity filters 111 have different parameters, providing different spectral transmittances in the long-wave infrared range (typically 7μm~12μm). The structure of the cavity filters 111 is fixed, i.e., the gas thickness (optical path = 5cm) is the same. Therefore, the different parameters of the N cavity filters 111 are controlled by the filling gas parameters. N is set to 6, and the filling gases are methane, ethane, propane, butane, ethylene, and propylene, all pure gases (100% concentration). In this embodiment, N=6, meaning the device contains six parallel spectral signal acquisition channels, but the invention is not limited to this; N can be flexibly adjusted according to factors such as spectral resolution, imaging field of view, and cost budget.

[0061] The infrared lens 112 employs a mid-infrared achromatic optical lens, such as a lens group made of zinc selenide or germanium. The infrared lenses 112 of each channel have the same or similar focal length and field of view to ensure spatial consistency of the images acquired by each channel.

[0062] Specifically, in this embodiment, the N infrared lenses 112 have identical parameters, are made of germanium material, and provide high spectral transmittance in the long-wave infrared range (typically 7μm~12μm). Above 95%.

[0063] The infrared detector 113 employs a mid-infrared focal plane array detector, such as a cooled infrared detector based on mercury cadmium telluride (MCT) or type-2 superlattice (T2SL) materials, or an uncooled vanadium oxide (VOx) microbolometer detector. The spectral response range of the detector should cover the mid-infrared band of the target.

[0064] Specifically, in this embodiment, N infrared detectors 113 have identical parameters, providing a high detection responsivity in the long-wave infrared range (typically 7μm~12μm). More than 70%.

[0065] The spatial registration unit 200 is connected to the array-type multispectral signal acquisition unit 100 and is used to perform spatial registration processing on the original multispectral signals acquired by N spectral signal acquisition units 110 to obtain the registered multispectral image signal. The spectral reconstruction unit 300 is connected to the spatial registration unit 200 and is used to perform spectral reconstruction based on the registered multispectral image signal to obtain the hyperspectral signal of the target scene.

[0066] In this embodiment, due to manufacturing tolerances and installation errors, the images from the N spectral signal acquisition units may have slight differences in the center position and scale of the field of view (typically on the sub-pixel level). The spatial registration unit 200 first extracts feature points from each channel image, using the SIFT (Scale Invariant Feature Transform) algorithm to detect and describe local feature points in the image; then, it performs feature matching and geometric transformation model estimation based on the RANSAC (Random Sample Consensus) algorithm to eliminate mismatched points; finally, it resamples each channel image through bicubic interpolation to achieve pixel-level precise alignment. After spatial registration processing, the same spatial target point corresponds to the same pixel coordinates in the N channel images, with a registration accuracy better than 0.5 pixels.

[0067] The spectral reconstruction unit 300 is connected to the spatial registration unit 200 and is used to perform spectral reconstruction based on the registered multispectral image signal to obtain the hyperspectral signal of the target scene.

[0068] In this embodiment, the spectral reconstruction unit is used to decouple and reconstruct the N-dimensional spectral channel signal into an S-dimensional spectral channel signal, which can be represented as follows: .in, This represents the spectral reconstruction operator, where S is the target number of hyperspectral channels. In this embodiment, N=6. =60, meaning that 60 channels of hyperspectral signal can be reconstructed from 6-dimensional spectral signal.

[0069] Preferably, the spectral reconstruction operator employs the Wiener estimation method, i.e.

[0070]

[0071] in, , Let be the autocorrelation matrix of the target hyperspectral signal. Let be the autocorrelation matrix of the noise.

[0072] System observation mapping matrix It can be represented as

[0073]

[0074] Where i is the multispectral channel number and j is the hyperspectral band number. For the corresponding wavelength interval, Invert a matrix.

[0075] Thus, this embodiment presents a mid-infrared spectral video computing camera device. It employs six cameras and a gas-cavity filter to construct an array-type multispectral signal acquisition unit. The gas-cavity filter uses methane, ethane, propane, butane, ethylene, and propylene gases as filling gases, providing different spectral transmittances. The device acquired multispectral signals in the 7μm~12μm wavelength range and ultimately reconstructed hyperspectral signals across 60 bands.

[0076] Example 3

[0077] like Figure 4 As shown, this embodiment provides a mid-infrared spectral video calculation camera method, including the following steps:

[0078] S10: N raw multispectral signals are simultaneously acquired by N spectral signal acquisition units of an array-type multispectral signal acquisition unit, where N is a positive integer greater than 1. Each spectral signal acquisition unit is composed of a cavity filter, an infrared lens, and an infrared detector coupled in sequence. The cavity filter is used to provide spectral transmittance in the target wavelength range and consists of a housing, optical windows respectively sealed at both ends of the housing, a filling gas, an inflation interface, and a sealing element. The spectral transmittance of the cavity filter can be adjusted by changing the parameters of the filling gas (including gas type, concentration, pressure, etc.).

[0079] In this embodiment, N=6, meaning that six spectral signal acquisition units are used to form an array-type multispectral signal acquisition unit. The housing of the gas cavity filter is a metal shell (such as aluminum alloy or stainless steel), and the optical window is an optical glass window. It is sealed to the metal shell by a sealing element (such as a rubber ring) to form a sealed cavity. The inflation port is located on the side wall of the metal shell and is used to fill or replace the filling gas in the sealed cavity.

[0080] Specifically, the original multispectral signal It is calculated using the following formula:

[0081]

[0082] in, The original multispectral signal, The original hyperspectral signal, For lens spectral transmittance, For the spectral responsivity of the infrared detector, This represents the spectral transmittance of the air-cavity filter.

[0083] S20: Perform spatial registration processing on the original multispectral signal to obtain a registered multispectral signal.

[0084] S30: Perform spectral decoupling and reconstruction on the registered multispectral signal to obtain the reconstructed hyperspectral signal.

[0085] In one possible implementation, the registered multispectral signal is calculated using the following formula:

[0086]

[0087] in, For registration of multispectral signals For spatial registration transformation operators.

[0088] In one possible implementation, the reconstructed hyperspectral signal is calculated using the following formula:

[0089]

[0090] Where H is the reconstructed hyperspectral signal, This represents the spectral reconstruction operator, where S is the number of target hyperspectral channels.

[0091] Preferably, in order to improve the accuracy of spectral acquisition and reconstruction of the array-type multispectral signal acquisition unit, before performing step S10, a preferred step S00 is included to pre-select the gas-filling filters for the N spectral signal acquisition units. A filling gas. For example... Figure 5 As shown, step S00 includes the following sub-steps:

[0092] S01: Select from the gas database according to preset rules A multispectral signal acquisition unit is constructed using gas samples. Each of the multispectral signal acquisition units corresponds to a selected gas sample, that is, the i-th multispectral signal acquisition unit is filled with the i-th selected gas sample, i=1,2,...,N.

[0093] In this embodiment, N=6; the gas database uses the gas spectral database published by NIST (National Institute of Standards and Technology), containing a total of 119 gas samples, numbered 1 to 119. The filling gas fills the sealed cavity, with the gas volume equal to the cavity volume, and the spectral transmittance of the filling gas is denoted as... , Therefore, the spectral transmittance of a cavity filter can be calculated using the following formula:

[0094]

[0095] in, The spectral transmittance of the air-cavity filter is [value missing]. The spectral transmittance of the optical window. The spectral transmittance of the filling gas. For wavelength, In this embodiment, the spectral transmittance of the optical window is approximately set to 0.95.

[0096] As described above, the spectral transmittance of a cavity filter is mainly determined by the combined spectral transmittance of the optical window and the filling gas. Typically, once the optical window is selected and the overall structure of the cavity filter is fabricated, the spectral transmittance is primarily determined by the spectral transmittance of the filling gas. Therefore, the spectral transmittance of the filter can be adjusted by changing the parameters of the filling gas (including its type, concentration, and pressure).

[0097] S02: The raw multispectral signal is acquired using an array-type multispectral signal acquisition unit. The raw multispectral signal is calculated using the following formula:

[0098]

[0099] in, The original multispectral signal, The original hyperspectral signal, This refers to the spectral transmittance of the infrared lens. This represents the spectral responsivity of the infrared detector.

[0100] In this embodiment, the ASTER spectral database was selected, containing 1500 measured ground object spectral data points, covering a wavelength range of 0.4. -15.4 It covers various land features including minerals, rocks, soil, vegetation, water bodies, man-made materials, and meteorites. The spectral transmittance of the infrared lens is approximately set to 0.95, and the spectral responsivity of the infrared detector is obtained from a table based on the selected detector model.

[0101] S03: Perform spatial registration on the original multispectral signal to obtain the registered multispectral signal. The registered multispectral signal is calculated using the following formula:

[0102]

[0103] in, Indicates the registered multispectral signal This represents the spatial registration transformation operator.

[0104] In this embodiment, since the ASTER spectral database is a single-point measurement result, and assuming that the original hyperspectral signal of the target scene is uniformly distributed in space, the signals obtained by each detector in the array-type multispectral signal acquisition unit can be regarded as the same target signal. Therefore, the spatial registration transformation operator is the standard identity matrix.

[0105] Spatial registration processing performs pixel-level spatial alignment of the original multispectral signal, which can be achieved using algorithms such as SIFT, SURF, or deep learning image registration algorithms.

[0106] S04: Perform spectral decoupling and reconstruction on the registered multispectral signal to obtain the reconstructed hyperspectral signal. The reconstructed hyperspectral signal is calculated using the following formula:

[0107]

[0108] In this embodiment, spectral decoupling reconstruction is used to... 3D spectral channel signal decoupling and reconstruction 3D spectral channel signal, N=6, =60, meaning that 60 channels of hyperspectral signal can be reconstructed from 6-dimensional spectral signal.

[0109] In this embodiment, the spectral reconstruction operator The Wiener estimation method is used, specifically expressed as follows:

[0110]

[0111] in, , Let be the autocorrelation matrix of the target hyperspectral signal. Let be the autocorrelation matrix of the noise.

[0112] System observation mapping matrix It can be represented as:

[0113]

[0114] Where i is the multispectral channel number and j is the hyperspectral band number. For the corresponding wavelength interval, Invert a matrix.

[0115] S05: Calculate the original hyperspectral signal and reconstructed hyperspectral signals Peak signal-to-noise ratio (PSNR).

[0116] In this embodiment, the hyperspectral signal uses the ASTER spectral database, which contains a total of 1500 samples, including 1000 training samples, to calculate the autocorrelation matrix of the target hyperspectral signal. The other 500 samples were used as a test set to calculate the peak signal-to-noise ratio (PSNR) between the reconstructed hyperspectral signal and the original hyperspectral signal.

[0117] S06: Repeat steps S01 to S05. Second-rate( Find the peak signal-to-noise ratio (PSNR) corresponding to the maximum value. Each gas sample was used as... The filling gas in a gas cavity filter.

[0118] In this embodiment, The process involves randomly selecting N=6 different gas samples from 119 gases in the NIST database as a candidate combination, and repeating this process 1000 times. In each iteration, six gas numbers (without repetition) are randomly selected to construct candidate array-type multispectral signal acquisition units. Then, steps S01 to S05 are executed sequentially to calculate the peak signal-to-noise ratio (PSNR) for each candidate combination. After 1000 iterations, the group of six gas samples that maximizes the PSNR is selected and assigned to six gas-cavity filters as filler gases.

[0119] In this embodiment, following the above calculation steps, the selected filling gas is, for example, gas samples numbered 8, 13, 23, 62, 78, and 105, whose spectral transmittance is as follows: Figure 6 As shown, when using this set of gas samples on the test set, the peak signal-to-noise ratio (PSNR) of the reconstructed hyperspectral signal is greater than 28.0 dB, indicating that the spectral reconstruction quality meets the requirements of high-precision imaging.

[0120] Thus, this embodiment presents a mid-infrared spectral video computation and imaging method. It employs six cameras and a gas-cavity filter to construct an array-type multispectral signal acquisition unit. Furthermore, the method for selecting the filling gas for the gas-cavity filter utilizes an iterative optimization step based on maximizing the peak signal-to-noise ratio (PSNR), ensuring high-precision spectral reconstruction. This device acquires multispectral signals within the 7μm~12μm band and ultimately reconstructs hyperspectral signals across 60 bands, achieving hyperspectral video imaging in the mid-infrared band.

[0121] It should be noted that the specific values ​​N=6, S=60, K=1000 in the above embodiments are merely examples. In practical applications, these values ​​can be flexibly adjusted according to factors such as spectral resolution requirements, target band range, and computational resource constraints. For example, in scenarios requiring higher spectral resolution, the value of N can be increased; when more thorough iterative optimization is needed, the value of K can be increased. However, the one-to-one correspondence between the N gas samples and the N spectral signal acquisition units should be maintained.

[0122] It should also be noted that the spectral reconstruction operator in the above embodiments Using Wiener's estimation method is only one preferred implementation. In other implementations, Alternatively, sparse reconstruction algorithms based on compressed sensing, data-driven reconstruction methods based on deep neural networks, or least-squares estimation methods based on pseudo-inverse matrices can be used, as long as they can effectively reconstruct N-dimensional multispectral signals into S-dimensional hyperspectral signals.

[0123] The above description is merely a preferred embodiment and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, all of which fall within the scope of protection of this application.

Claims

1. A cavity-type optical filter, characterized in that, include: The housing has optical window mounting positions at both ends; Two optical windows are respectively sealed and installed at the optical window mounting positions at both ends of the housing, forming a sealed cavity with the housing; An inflation port is provided on the housing and communicates with the sealed cavity for filling the sealed cavity with filling gas.

2. The air-cavity filter according to claim 1, characterized in that, The housing is a metal shell; the optical window is an optical glass window; a seal is provided at the connection between the optical glass window and the metal shell; the inflation port is located on the side wall of the metal shell.

3. A mid-infrared spectral video computing camera device, characterized in that, The device includes: An array-type multispectral signal acquisition unit is used to acquire raw multispectral signals. The array-type multispectral signal acquisition unit includes N spectral signal acquisition units, where N is a positive integer greater than 1. Each spectral signal acquisition unit comprises a cavity filter, an infrared lens, and an infrared detector arranged sequentially along the optical axis. The cavity filter includes a housing, optical windows sealed at both ends of the housing, and an inflation port on the housing that communicates with the sealed cavity. The optical windows and the housing form the sealed cavity, and the inflation port is used to fill the sealed cavity with a filling gas. The cavity filter provides spectral transmittance within a target wavelength range, and the spectral transmittance of the cavity filter is adjusted by changing the parameters of the filling gas. The spatial registration unit is used to perform spatial registration processing on the original multispectral signal to obtain the registered multispectral signal. The spectral reconstruction unit is used to perform spectral reconstruction processing on the registered multispectral signal to obtain the reconstructed hyperspectral signal.

4. The mid-infrared spectral video computing camera device according to claim 3, characterized in that, The housing is a metal shell; the optical window is an optical glass window; a seal is provided at the connection between the optical glass window and the metal shell; the inflation port is located on the side wall of the metal shell.

5. A mid-infrared spectral video calculation camera method, implemented based on the mid-infrared spectral video calculation camera device according to any one of claims 3-4, characterized in that, The method includes: S10 acquires N raw multispectral signals through N spectral signal acquisition units of the array-type multispectral signal acquisition unit; S20, spatial registration processing is performed on the original multispectral signal to obtain the registered multispectral signal; S30 performs spectral reconstruction processing on the registered multispectral signal to obtain the reconstructed hyperspectral signal.

6. The mid-infrared spectral video calculation and imaging method according to claim 5, characterized in that, It also includes step S00, which involves pre-selecting the air-cavity filters for the N spectral signal acquisition units. One filling gas, specifically including: S01: Select from the gas database according to preset rules A gas sample was used to construct an array-type multispectral signal acquisition unit; S02: Acquire the raw multispectral signal through an array-type multispectral signal acquisition unit; S03: Perform spatial registration processing on the original multispectral signal to obtain the registered multispectral signal; S04: Perform spectral reconstruction processing on the registered multispectral signal to obtain the reconstructed hyperspectral signal; S05: Calculate the peak signal-to-noise ratio of the original hyperspectral signal and the reconstructed hyperspectral signal; S06: Repeat steps S01 to S05. Second-rate, It is a positive integer. Find the peak signal-to-noise ratio corresponding to the maximum value. Each gas sample was used as... The filling gas in a gas cavity filter.

7. The mid-infrared spectral video calculation and imaging method according to claim 5, characterized in that, The spectral transmittance of the air cavity filter is calculated using the following formula: ； in, The spectral transmittance of the air-cavity filter is [value missing]. The spectral transmittance of the optical window. The spectral transmittance of the filling gas, For wavelength, .

8. The mid-infrared spectral video calculation and imaging method according to claim 7, characterized in that, The original multispectral signal is calculated using the following formula: ； in, The original multispectral signal, The original hyperspectral signal, For lens spectral transmittance, denoted as the detector's spectral responsivity.

9. The mid-infrared spectral video calculation and imaging method according to claim 8, characterized in that, The registered multispectral signal is calculated using the following formula: ； in, For the registered multispectral signal For spatial registration transformation operators.

10. The mid-infrared spectral video calculation and imaging method according to claim 9, characterized in that, The reconstructed hyperspectral signal is calculated using the following formula: ； Where H is the reconstructed hyperspectral signal, Here, S is the spectral reconstruction operator, and S is the number of target hyperspectral channels.

Images