A space-based infrared hyperspectral imaging index design method

Abstract

In order to ensure the efficient application of the infrared hyperspectral load in orbit, a space-based infrared hyperspectral imaging index design method is provided. Step one, according to the satellite detection task requirement, the minimum joint signal-to-clutter ratio of the corresponding infrared hyperspectral image is calculated; Step two, according to the satellite application requirement, the infrared hyperspectral radiance image of the task target scene is generated; Step three, according to the radiance signal of the task target scene, the corresponding space-based infrared hyperspectral image is simulated and generated by using the infrared hyperspectral radiance image generated in step two; Step four, based on the characteristic spectrum in the space-based infrared hyperspectral image, the optimal joint signal-to-clutter ratio of the infrared hyperspectral image is calculated; Step five, the relationship between the joint signal-to-clutter ratio and the imaging index is established, and the influence weight of each index is analyzed; Step six, based on the relationship and the influence weight obtained in step five, an optimization function is constructed, the infrared hyperspectral imaging index is optimized through the optimization function, and the optimized design index is obtained.

Description

Technical Field

[0001] This invention relates to a method for designing space-based infrared hyperspectral imaging parameters, which is mainly applied to the field of infrared hyperspectral satellite payload parameter design. Background Technology

[0002] Infrared radiation contains objective information not found in the visible light band. Particularly in imaging, infrared imaging systems possess the unique advantage of operating around the clock, and have gained significant attention from various countries for their application in aerospace remote sensing. Compared to traditional single-band / multi-band infrared imaging techniques, infrared hyperspectral detection can obtain spectral radiation information from many consecutive narrow bands, enabling precise detection of the spectral distribution characteristics of targets and significantly improving target detection and identification capabilities. It is especially effective in detecting targets with very low contrast to the background radiation, playing an irreplaceable role in space-based detection of sea, land, and air targets.

[0003] Imaging performance design is a prerequisite for the design of infrared hyperspectral imaging systems and a crucial foundation for ensuring the efficient on-orbit application of infrared hyperspectral remote sensing imaging. Unlike the visible light band, infrared targets exhibit flat emissivity curves and insignificant spectral characteristics. Weak spectral features are difficult to detect and characterize, and even more difficult to separate from atmospheric effects. Therefore, spectral detection places higher demands on indicators such as noise equivalent temperature difference (NETD). Furthermore, space-based infrared hyperspectral detection is limited by various factors across the entire process, including weak infrared radiation energy, the significant impact of water vapor on transmittance, the high difficulty in developing infrared detectors, and limited calibration accuracy. This results in a discrepancy between the actual application level of infrared hyperspectral detection and theoretical analysis. Therefore, it is necessary to establish a design method for space-based infrared hyperspectral imaging performance indicators that considers the correlation between performance indicators under non-ideal conditions and practical applications, establish the relationship between imaging performance indicators and application performance, and guide the design of infrared hyperspectral imaging systems.

[0004] Research on the design of infrared hyperspectral imaging indicators for target detection applications is relatively lagging. In the 1990s, Eismann first proposed the concept of joint signal-to-noise ratio in the United States and conducted targeted research on target detection during hot crossover periods. Among them, "Comparison of infrared imaging hyperspectral sensors for military target detection applications" (SPIE, 1996) analyzed the detection capability of infrared hyperspectral imaging for targets during hot crossover periods. However, this work was mainly aimed at ground applications and did not consider the influence of space-based observation conditions such as the atmosphere. At the same time, this work only analyzed the requirements of ground applications for relevant indicators and did not analyze the optimization strategies of relevant indicators, so it could not form a complete design method. Zhao Huijie of Beijing University of Aeronautics and Astronautics published "Application of Infrared Multispectral Technology in Detection During Day-Night Transition Periods" (Infrared and Laser Engineering, 2018, 47(2)), which proposed that infrared multispectral technology can enhance the target detection capability under low target background contrast conditions during hot crossover periods and provided the design of an infrared multispectral detection system. However, this system only contains 5 infrared bands and will not be affected by the limiting factors such as energy radiation introduced by hyperspectral imaging and detector development level. Also, it did not consider the influence of space-based observation conditions such as the atmosphere.

[0005] In summary, there is currently no index design method based on full-link application performance analysis for space-based infrared hyperspectral imaging to ensure that the infrared hyperspectral imaging system meets the actual on-orbit application requirements of the satellite. This is also an important prerequisite and challenge in the design of infrared hyperspectral satellites. Summary of the Invention

[0006] To ensure the efficient application of infrared hyperspectral payloads in orbit, this invention proposes a method for designing space-based infrared hyperspectral imaging parameters.

[0007] The present invention is achieved through the following technical solution.

[0008] A method for designing space-based infrared hyperspectral imaging parameters includes the following steps:

[0009] Step 1: Calculate the lowest joint signal-to-noise ratio for the corresponding infrared hyperspectral image according to the requirements of the satellite detection mission;

[0010] Step 2: Generate infrared hyperspectral radiance images of the mission target scene according to satellite application requirements;

[0011] Step 3: Based on the radiance signal of the target scene, use the infrared hyperspectral radiance image generated in Step 2 to simulate and generate the corresponding space-based infrared hyperspectral image;

[0012] Step 4: Based on the characteristic spectral bands in the space-based infrared hyperspectral image, calculate the optimal joint signal-to-noise ratio of the infrared hyperspectral image;

[0013] Step 5: Repeat steps 3 and 4 to establish the relationship between the joint signal-to-clutter ratio and imaging parameters, and analyze the influence weight of each parameter;

[0014] Step 6: Construct an optimization function based on the relationships and influence weights obtained in Step 5, and use this optimization function to optimize the infrared hyperspectral imaging index to obtain the optimized design index.

[0015] The beneficial effects of this invention are:

[0016] 1. Based on the satellite's detection rate / false alarm rate requirements for typical scenes and targets, this invention calculates the joint signal-to-noise ratio (SNR) threshold requirement for hyperspectral images. Simultaneously, through full-link simulation, it obtains space-based infrared hyperspectral images under the corresponding indicators and calculates the SNR, establishes the relationship between imaging indicators and SNR, analyzes the influence weight of each indicator on SNR, and optimizes imaging indicators based on the SNR threshold and indicator influence weight.

[0017] 2. This invention is applicable to the optimization design of infrared hyperspectral imaging indicators for target detection applications. It evaluates the effectiveness of target detection applications by combining the image and spectrum with the signal-to-noise ratio. It considers the impact of all factors in the entire link from target characteristics to atmosphere to space-based imaging system on target detection applications, realistically simulates the on-orbit imaging state, and confirms the optimization weight of indicators through sensitivity analysis to ensure the optimal indicator design for real space-based applications.

[0018] 3. This invention employs an end-to-end infrared hyperspectral full-link simulation method, from target characteristics to atmosphere to space-based imaging system, to realistically simulate the on-orbit imaging state;

[0019] 4. This invention optimizes the indicators by combining the signal-to-noise ratio of the image and simultaneously correlates the imaging performance indicators with the application indicators. The optimization method is simple and computationally efficient. It can not only accurately evaluate the application capability of the infrared hyperspectral payload, but also enable the design of imaging indicators to be oriented towards real on-orbit applications.

[0020] 5. This invention obtains the influence weights of different indicators on target detection application indicators through sensitivity analysis based on the random forest model, and introduces the influence weights into the process of imaging indicator optimization design to ensure that the indicator optimization process is more reasonable, effectively avoid the situation of excessively high related indicators, and reduce the development cost of payload. Attached Figure Description

[0021] Figure 1 This is a flowchart of the infrared hyperspectral imaging index design method of the present invention;

[0022] Figure 2This is a graph showing the relationship between the combined signal-to-noise ratio and the target detection rate / false alarm rate of this invention. Detailed Implementation

[0023] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the present invention, and are not intended to limit the scope of the present invention.

[0024] The principle of this invention is based on the satellite's detection rate / false alarm rate requirements for typical scenes and targets, which allows for the calculation of the joint signal-to-noise ratio (SNR) threshold requirement for hyperspectral images. Simultaneously, through full-link simulation, space-based infrared hyperspectral images under the corresponding indicators are obtained and the SNR is calculated. The relationship between imaging indicators and SNR is established, and the influence weight of each indicator on SNR is analyzed. Based on the SNR threshold and the influence weight of the indicators, imaging indicators are optimized.

[0025] like Figure 1 As shown, the method for designing space-based infrared hyperspectral imaging indicators according to the present invention specifically includes the following steps:

[0026] Step 1: Calculate the lowest joint signal-to-noise ratio (SCR) of the corresponding infrared hyperspectral image according to the satellite detection mission requirements. th ;

[0027] In this embodiment, the satellite detection task requires a detection rate P. d0 and false alarm rate P f0 The specific relationship between the detection rate, false alarm rate, and joint signal-to-clutter ratio is as follows:

[0028] P D =1-Q(SCR-Q) -1 (P F ))

[0029] Among them, P F Let Q be the constant false alarm rate, and Q be the right-tail function of the standard normal distribution, which is calculated as follows:

[0030]

[0031] like Figure 2 As shown, substituting different false alarm rates into the above two equations yields the curve relationship between target detection rate, false alarm rate, and image joint signal-to-clutter ratio. Calculations show that when the false alarm rate is 10... -7 When, SCR satisfies SCR low When the target detection rate is ≥6.4, it meets the requirements of the satellite mission.

[0032] In practice, to avoid excessively high payload development costs due to overly ambitious specifications, the upper limit of SCR is set at SCR. high ~SCR low+1, the threshold for the image joint signal-to-noise ratio is set to SCR. th =7.5; This approach ensures that subsequent performance optimization meets application requirements while avoiding excessive payload development costs and excessively high performance indicators.

[0033] Step 2: Generate an infrared hyperspectral radiance image of the mission target scene according to satellite application requirements; specifically, the following method is used:

[0034] Based on the requirements of satellite applications, the target scenario is determined, and it is determined whether the emissivity and reflectivity of the target scenario are known. If so, the temperatures of the target and background are set, and an infrared hyperspectral radiance image is obtained through simulation. Otherwise, ground or airborne tests are conducted, and a hyperspectral image is obtained through a high-performance thermal infrared hyperspectral detector as the infrared hyperspectral radiance image of the target scenario.

[0035] In this embodiment, the infrared hyperspectral radiance image obtained through simulation is specifically calculated using the following formula:

[0036] L0(x0,y0,λ)=L e (x0,y0,λ)+L r (x0,y0,λ)

[0037] Among them, L e (x0, y0, λ) represents the radiation spectrum signal of the target scene, L r (x0,y0,λ) represents the reflectance spectral signal of the target scene, (x0,y0) represents the coordinates of the target scene image, and λ represents the wavelength.

[0038] The radiation spectrum signal of the target scene is as follows:

[0039]

[0040] Where ε(x0,y0,λ) is the emissivity of the target / background, T is the set temperature, h is Planck's constant, c is the speed of light, and k is Boltzmann's constant;

[0041] The reflectance spectral signal of the target scene is as follows:

[0042]

[0043] Where ρ(x0,y0,λ) is the target / background reflectivity, E0(λ) is the energy of the sun directly incident on the target / background after passing through the atmosphere, which can be obtained through MODTRAN simulation, and θ s L is the solar zenith angle. s-down The downward solar radiation can be obtained through MODTRAN simulation. In practice, when the imaging time is nighttime, L r(x0,y0,λ)=0.

[0044] Step 3: Based on the radiance signal of the target scene, use the infrared hyperspectral radiance image generated in Step 2 to simulate and generate the corresponding space-based infrared hyperspectral image; the specific steps are as follows:

[0045] S301. The radiance signal of the mission target scene first reaches the entrance pupil of the space-based infrared hyperspectral payload after atmospheric transmission. The radiance signal at the entrance pupil is:

[0046] L1(x0,y0,λ)=τ(λ)L0(x0,y0,λ)+L p (λ)+L s-up (λ)

[0047] Where τ(λ) is the atmospheric spectral transmittance, which can be obtained through MODTRAN simulation; L p (λ) represents the radiance of the atmospheric path radiation, which can be obtained through MODTRAN simulation; L s-up The radiation emitted from the sun can be obtained through MODTRAN simulation.

[0048] S302. Using the radiance signal at the entrance pupil as input, and based on the principle of infrared hyperspectral imaging, simulate and obtain a space-based infrared hyperspectral image under the infrared hyperspectral imaging index.

[0049] In this embodiment, the simulation mainly includes spatial response simulation, spectral response simulation, and radiation response simulation; the specific steps are as follows:

[0050] Spatial response simulation involves bilinear interpolation spatial resampling of the entrance pupil radiance image based on spatial resolution, and convolution of the image based on the system MTF-simulated point spread function (PSF) to obtain the spatially responded infrared hyperspectral image; the formula is as follows:

[0051]

[0052] Where (x,y) are the influence coordinates of (x0,y0) after spatial resolution resampling of infrared hyperspectral imaging, and the point spread function (PSF) is simulated using a two-dimensional Gaussian function, with its half-width and full height (FWHM) as follows:

[0053] FWHM=7*σ psf

[0054] in,

[0055] Spectral response simulation obtains the radiance signal for each band by convolving the infrared hyperspectral image with the spectral response function; the formula is as follows:

[0056]

[0057] Where, λ i The SRF is the spectral response function obtained after resampling according to the spectral resolution, which can be approximated by a Gaussian function with a spectral resolution of full width at half maximum (FWHM).

[0058] The radiation response simulation, based on the noise equivalent radiance corresponding to NETD and the random multiplicative error introduced by radiometric calibration, ultimately yields the space-based infrared hyperspectral image; the formula is as follows:

[0059] L'(x,y,λ i )=(L3(x,y,λ i )+randn*NESR)*e k

[0060] Wherein, NESR is the noise equivalent temperature difference, NESR=B(T+NETD,λ)-B(T,λ), and B(*) is the Planck blackbody radiation calculation formula;

[0061] Among them, e k e is the random multiplicative error coefficient. k =rand*e rad / 2+1,e rad This refers to the absolute calibration error;

[0062] Where randn is Gaussian random noise with a mean of 0.

[0063] Step 4: Based on the characteristic spectral bands in the space-based infrared hyperspectral image, calculate the optimal joint signal-to-noise ratio of the infrared hyperspectral image;

[0064] The formula for the optimal joint signal-to-noise ratio is as follows:

[0065]

[0066] Where s is the contrast vector between the target spectrum and the background spectrum, the length of the vector is the number of preferred spectral segments N, and M is the covariance matrix of the background, the width of which is equal to the number of preferred spectral segments N.

[0067] When N≤2, the spectral segment with the best joint signal-to-noise ratio is selected by traversing the spectral segments; when N>2, based on the N-1 spectral segments, the remaining spectral segments are traversed and combined with the first N-1 spectral segments to form N spectral segments and the joint signal-to-noise ratio is calculated. The N spectral segments with the highest joint signal-to-noise ratio are selected as the preferred N spectral segments, and the calculated joint signal-to-noise ratio is taken as the optimal joint signal-to-noise ratio.

[0068] Step 5: Repeat steps 3 and 4 to establish the relationship between the joint signal-to-clutter ratio and imaging parameters, and analyze the influence weight of each parameter;

[0069] In this embodiment, the relationship between the joint signal-to-noise ratio and imaging parameters is specifically expressed by the following formula:

[0070] SCR var =f(var)

[0071] Here, var represents three main indicators affecting infrared hyperspectral imaging: spatial resolution Δd, spectral resolution Δλ, and NETDΔT. Keeping two of these indicators constant and changing the third, the corresponding joint signal-to-clutter ratio (SCR) of the hyperspectral image is obtained. Simultaneously, the SCR calculated under various combinations of indicators is used as a matrix input to a random forest model, and deep learning is used to obtain the influence weights of each indicator on the joint SCR.

[0072] {R var}=RF{Δd,Δλ,ΔT,SCR}

[0073] Among them, R var The weights of the three indicators represent their influence, and satisfy the following conditions: RF stands for Random Forest, a deep learning model.

[0074] Step 6: Construct an optimization function based on the relationships and influence weights obtained in Step 5, and use this optimization function to optimize the infrared hyperspectral imaging index to obtain the optimized design index;

[0075] In this embodiment, the specific formula of the optimization function is as follows:

[0076]

[0077] stvar min ≤var≤var max

[0078] Where, var max var represents the design requirements for satellite performance. min This represents the optimal value achievable based on the current level of research and development.

[0079] In practice, after obtaining the optimized design indicators in step six, the indicators are further verified to ensure that they meet the application requirements; specifically, the following methods are used:

[0080] The optimized design index obtained in step six is ​​input back into step three, and the joint signal-to-clutter ratio of infrared hyperspectral images under the optimized index is obtained through step four. The false alarm rate P is then calculated. f0 The target detection rate P d Determine P d ≥P d0 If the conditions are met, output the optimization index result; otherwise, return to step six to optimize again.

[0081] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for designing a space-based infrared hyperspectral imaging index, characterized in that, Includes the following steps: Step 1: Calculate the minimum joint signal-to-noise ratio (SNR) of the corresponding infrared hyperspectral image according to the satellite detection mission requirements; the satellite detection mission requirements include the detection rate and the false alarm rate. Step 2: Generate infrared hyperspectral radiance images of the mission target scene according to satellite application requirements; Step 3: Based on the radiance signal of the target scene, use the infrared hyperspectral radiance image generated in Step 2 to simulate and generate the corresponding space-based infrared hyperspectral image; Step 4: Based on the characteristic spectral bands in the space-based infrared hyperspectral image, calculate the optimal joint signal-to-noise ratio of the infrared hyperspectral image; Step 5: Repeat steps 3 and 4 to establish the relationship between the joint signal-to-clutter ratio and imaging parameters, and analyze the influence weight of each parameter; the specific formula for the relationship between the joint signal-to-clutter ratio and imaging parameters is as follows: in, Three main indicators affecting infrared hyperspectral imaging: spatial resolution Spectral resolution and NETD Keeping two of the indicators constant and changing the other indicator, we obtain the corresponding joint signal-to-noise ratio (SNR) of the hyperspectral image. At the same time, we input the SNR calculated under the combination of various indicators as a matrix into the random forest model, and obtain the influence weight of each indicator on the joint SNR through deep learning. Step 6: Construct an optimization function based on the relationships and influence weights obtained in Step 5, and use this optimization function to optimize the infrared hyperspectral imaging index to obtain the optimized design index; The specific formula for the optimization function is as follows: in, These are the design requirements for satellite specifications. This represents the optimal value achievable based on the current level of research and development. This represents the influence weight of the three indicators.

2. The method of claim 1, wherein, Step two is specifically implemented in the following manner: Based on the requirements of satellite applications, the target scenario is determined, and it is determined whether the emissivity and reflectivity of the target scenario are known. If so, the temperatures of the target and background are set, and an infrared hyperspectral radiance image is obtained through simulation. Otherwise, ground or airborne tests are conducted, and a hyperspectral image is obtained through a high-performance thermal infrared hyperspectral detector as the infrared hyperspectral radiance image of the target scenario.

3. The method of claim 2, wherein: The infrared hyperspectral radiance image is obtained through simulation, and the specific formula is as follows: in, The radiation spectrum signal of the target scene. The reflected spectrum signal of the target scene. The coordinates of the scene image of the mission target. Wavelength; The radiation spectrum signal of the target scene is as follows: in, Emittance of target / background The set temperature, Let be Planck's constant. At the speed of light, Boltzmann's constant; The reflectance spectral signal of the target scene is as follows: wherein, is the target / background reflectivity, is the solar energy directly incident on the target / background after passing through the atmosphere, is the solar zenith angle, is the solar downwelling radiation.

4. The method for designing space-based infrared hyperspectral imaging parameters as described in claim 1 or 2, characterized in that, Step 3 is detailed below: S301. The radiance signal of the mission target scene first reaches the entrance pupil of the space-based infrared hyperspectral payload through atmospheric transmission. S302. Using the radiance signal at the entrance pupil as input, and based on the principle of infrared hyperspectral imaging, simulate and obtain a space-based infrared hyperspectral image under the infrared hyperspectral imaging index.

5. The method of claim 4, wherein: The simulation includes spatial response simulation, spectral response simulation, and radiation response simulation; the specific steps are as follows: The spatial response simulation will perform bilinear interpolation spatial resampling on the entrance pupil radiance image based on spatial resolution, and convolve the image based on the system MTF to simulate the point spread function PSF, so as to obtain the spatially responded infrared hyperspectral image. The spectral response simulation obtains the radiance signal of each band by convolving the infrared hyperspectral image with the spectral response function; The radiation response simulation, based on the noise equivalent radiance corresponding to NETD and the random multiplicative error caused by radiometric calibration, finally yields the space-based infrared hyperspectral image.

6. The method of claim 5, wherein: The formula for the optimal joint signal-to-noise ratio mentioned in step four is as follows: wherein is a contrast vector of the target spectrum with the background spectrum, the length of the vector being the number of preferred spectral bands N, is a covariance matrix of the background, the width of the matrix being equal to the number of preferred spectral bands N; When N≤2, the spectral segments with the best joint signal-to-noise ratio are traversed through the spectral segments; when N>2, based on N-1 spectral segments, the remaining spectral segments are traversed and combined with the first N-1 spectral segments to form N spectral segments, and the joint signal-to-noise ratio is calculated. The N spectral segments with the highest joint signal-to-noise ratio are selected as the preferred number of spectral segments, and the calculated joint signal-to-noise ratio is taken as the optimal joint signal-to-noise ratio.

7. The method of claim 5, wherein the method further comprises: After obtaining the optimization design index in step six, the index is verified to ensure that the optimization index meets the application index requirements; the following methods are adopted: The optimized design parameters obtained in step six are input back into step three, and the joint signal-to-noise ratio of infrared hyperspectral images under the optimized parameters is obtained through step four. The false alarm rate is then calculated. Target detection rate ,judge If the conditions are met, output the optimization index result; otherwise, return to step six to optimize again.

Images