A method for designing an RGB filter for low-light-level night vision

Abstract

The application discloses a design method of an RGB filter for low-light-level night vision, and belongs to the technical field of low-light-level night vision imaging. Data acquisition and preprocessing are carried out to obtain a spectral electron flux curve; a filter bandpass curve and process parameters are set; channel performance evaluation is carried out according to the spectral electron flux curve and the process parameters; multi-objective optimization and process adaptation are carried out through a two-step optimization strategy of global search and local fine adjustment; and channel calibration and enhancement are carried out to obtain final RGB filter transmittance data. The application aims to solve the technical problems of low light transmittance efficiency, large channel interference and poor manufacturing feasibility of the RGB filter in a low-light environment; improve total flux, channel energy balance, channel spectral overlap suppression and filter curve manufacturability; and introduce a conditional enhancement strategy for the B channel to make up for the short board of insufficient response of the EBCMOS in the B light band.

Description

Technical Field

[0001] This invention belongs to the field of low-light night vision imaging technology, specifically relating to a design method for RGB filters for low-light night vision. Background Technology

[0002] In low-light imaging systems, the end-to-end sensitivity of electron-bombarded CMOS (EBCMOS) is limited by the photocathode spectral response, input / sealing window transmittance, and readout chain noise. Different photocathode materials exhibit varying spectral cutoff characteristics; among them, GaAs photocathodes, for example, possess high quantum efficiency in the short-wave near-infrared, making the effective energy available in this band more readily available in nighttime environments. Traditional methods relying on adjusting local parameters such as center wavelength and bandwidth struggle to balance the global trade-offs of flux, overlap, and manufacturability, and fail to systematically explore the potential of combined VIS-SWNIR utilization, easily leading to problems such as excessive back-end white balance gain, color noise amplification, and unstable demixing.

[0003] In the prior art, the solutions closest to this invention mainly fall into three categories: one is to set the center wavelength and bandwidth of the filter according to the fixed RGB banding principle of visible light; another is to optimize the passband position based on the target reflectance spectrum, standard color chart, or color space constraints; and the third is to perform empirical parameter tuning only around local parameters such as bandwidth and center wavelength. These solutions generally lack an integrated indicator system and optimization framework that starts from the complete link of "environmental spectrum—device quantum efficiency—filter transmittance—system response," and in particular, they do not fully utilize the high response characteristics of GaAs photocathodes in the VIS-SWNIR band. Therefore, under low-light conditions, problems such as insufficient total flux, imbalanced channel energy ratio, excessive channel overlap, and difficulty in simultaneously meeting manufacturing constraints are prone to occur.

[0004] This invention targets an electron-bombarded CMOS (EBCMOS) detector chain employing GaAs photocathodes, proposing a design method, device, and storage medium for a visible-short-wave near-infrared (VIS–SWNIR) three-channel filter for low-light night vision. The three channels cover the visible light band and extend into the short-wave near-infrared, fully utilizing the radiant energy of the nighttime spectrum in the short-wave near-infrared and the high quantum efficiency of the GaAs photocathode in this band under extremely low illumination conditions. This improves the overall system throughput, reduces channel spectral overlap, and enhances color reproduction stability and imaging signal-to-noise ratio (SNR) while meeting manufacturability constraints. Summary of the Invention

[0005] The purpose of this invention is to provide a method for designing filter transmittance curves, given only the ambient light spectrum and the end-to-end equivalent quantum efficiency of GaAs-EBCMOS as input parameters. This method constructs a unified spectral electron flux curve and system response model, jointly optimizing the spectral transmittance curves of the visible light-shortwave near-infrared (VIS-SWNIR) three channels. This invention balances improvements in total flux, channel energy balancing, suppression of spectral overlap between channels, and manufacturability of the filter curve morphology. Furthermore, it introduces a conditional enhancement strategy for the B-color channel to compensate for the insufficient response of EBCMOS in the B-band.

[0006] This invention discloses a design method for RGB filters for low-light night vision, belonging to the field of low-light night vision imaging technology. The method involves data acquisition and preprocessing to obtain spectral electron flux curves; setting filter bandpass curves and process parameters; evaluating channel performance based on the spectral electron flux curves and process parameters; performing multi-objective optimization and process adaptation through a two-step optimization strategy of global search and local fine-tuning; and obtaining the final RGB filter transmittance data after channel calibration and enhancement. This invention aims to solve the technical problems of low light transmission efficiency, high channel interference, and poor manufacturing feasibility of RGB filters in low-light environments. It improves the manufacturability of total flux, channel energy balancing, spectral overlap suppression between channels, and filter curve morphology, and introduces a conditional enhancement strategy for the B-color channel to compensate for the insufficient response of EBCMOS in the B-band.

[0007] A design method for RGB filters for low-light night vision, such as Figure 1 As shown, it includes the following steps:

[0008] S1, acquire environmental spectral data and detector quantum efficiency function, perform peak normalization processing at a unified wavelength sampling point, and construct a normalized spectral electronic characterization function by combining the metric domain conversion factor;

[0009] S2, establish a parameterized expression for the transmittance function of a three-channel filter, and construct a family of manufacturable three-channel filter transmittance functions in combination with constraints; the constraints include process manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints;

[0010] S3. Multiply the normalized spectral electronic characterization function with the transmittance function of the three-channel filter to construct the nominal response function of the three channels, and calculate the single-channel relative flux, the total relative flux of the three channels, the channel energy ratio, the channel correlation evaluation quantity and the overall separability evaluation quantity of the three channels.

[0011] S4. Construct a comprehensive optimization objective function; under the constraints obtained in S2, solve for the comprehensive optimization of the transmittance function parameters of the three-channel filter.

[0012] S5 sequentially performs peak limiting calibration, B-channel energy discrimination, and limiting compensation processing on the optimized three-channel filter transmittance function output from S4 to obtain the final RGB filter transmittance data that meets the requirements of film system realization and batch consistency.

[0013] Furthermore, in S1, the process of constructing the normalized spectral electronic characterization function includes:

[0014] The environmental spectral data refers to the environmental spectral distribution under nighttime light conditions or darkroom conditions. ;

[0015] The quantum efficiency function of the detector is the quantum efficiency function of the electron-bombarded CMOS detector chain using gallium arsenide (GaAs) photocathode as the photosensitive material at different wavelengths. ;

[0016] The environmental spectral distribution With quantum efficiency function All in wavelength As the independent variable, it is used to characterize the relationship between the external radiation input and the quantum efficiency characteristics of the detector under low light conditions in the wavelength dimension;

[0017] Step 1.1, analyze the environmental spectral distribution. With quantum efficiency function Mapping to a unified wavelength sampling point and defining a unified operating band. ;

[0018] The mapping is the mapping of the environmental spectral distribution. With quantum efficiency function In a unified working band Resampling is performed internally to make the two correspond to the same set of wavelength sampling points, which are then used as input for subsequent calculations.

[0019] Step 1.1.1, analyze the environmental spectral distribution. With quantum efficiency function The original discrete data is subjected to interpolation and noise reduction processing;

[0020] Step 1.1.2: Select the band according to the preset unified working band; unified working band Determined by the spectral range of the detector's quantum efficiency;

[0021] Furthermore, the unified operating band It is limited by both the cutoff wavelength of the GaAs photocathode and the effective transmission range of the system's transmission window;

[0022] Step 1.2, map the environmental spectral distribution and quantum efficiency function Peak value normalization was performed separately:

[0023] Based on environmental spectral distribution In a unified working band Maximum value and quantum efficiency function In a unified working band The maximum value within the range is used as the normalization benchmark for the mapped environmental spectral distribution. and quantum efficiency function Amplitude scaling is performed to obtain the environmental spectral distribution after peak normalization. and the quantum efficiency function after peak normalization :

[0024] ;

[0025] ;

[0026] Where λ is the wavelength. and The maximum value of each is 1, thus preserving the original spectral characteristics and eliminating amplitude scale differences;

[0027] Step 1.3: Construct an initial spectral electronic characterization function at a uniform wavelength sampling point to obtain a normalized spectral electronic characterization function;

[0028] Step 1.3.1: Under the selected metric domain, normalize the environmental spectral distribution of the peak values. Quantum efficiency function after peak normalization and metric domain conversion factor Multiplication yields the initial spectral electronic characterization function. :

[0029] ;

[0030] The initial spectral electronic characterization function is used to characterize the initial relative spectral distribution result formed under the combined effect of environmental spectral distribution, quantum efficiency function and metric domain transformation relationship;

[0031] The metric domain conversion factor Used to characterize the spectral quantity conversion relationship under different measurement methods;

[0032] When using power domain representation ;

[0033] When using photon number field representation ;

[0034] Where h is Planck's constant and c is the speed of light in a vacuum;

[0035] Step 1.3.2, characterize the initial spectral electronic function. After peak normalization, the normalized spectral electronic characterization function is obtained. :

[0036] ;

[0037] in, satisfy: .

[0038] Furthermore, the initial spectral electronic characterization function is constructed using the photon number domain. and normalized spectral electronic characterization function This is to enhance the consistency of spectral shape comparison under different illumination conditions and different scene conditions, and to improve the stability of the subsequent filter transmittance function design process;

[0039] The normalized spectral electronic characterization function This is used to characterize the relative spectral distribution characteristics formed by the combined effect of the environmental spectral distribution and the detector quantum efficiency function at a unified wavelength sampling point, after peak normalization. It serves as the spectral characterization basis for the subsequent construction of the three-channel nominal response function and the comprehensive optimization solution.

[0040] Furthermore, in S2, the process of constructing the transmittance function of the three-channel filter is as follows:

[0041] The three-channel filter includes an R channel, a G channel, and a B channel, with channel indices denoted as follows: ;

[0042] The R, G, and B are only used to distinguish the three filter channel identifiers and do not limit their passbands to correspond to the red, green, and blue bands in the standard visual sense.

[0043] The transmittance function of the i-th channel filter is denoted as: Characterizes the wavelength of the i-th channel filter. The transmittance at a given point as a function of wavelength;

[0044] The transmittance function of the three-channel filter , and These will serve as the design objects for the subsequent construction and comprehensive optimization of the three-channel nominal response functions;

[0045] Step 2.1: Select the parameterized bandpass function family of the three-channel filter transmittance function;

[0046] Transmittance function of a three-channel filter Parametric characterization is performed using symmetric superGaussian bandpass functions:

[0047] ;

[0048] in, The center wavelength of the i-th channel is... Let be the peak transmittance of the i-th channel. n is the scale parameter. i Shape order;

[0049] The shape order n i Used to adjust the flatness of the top of the transmittance function and the steepness of the transition at the edge of the passband;

[0050] Each channel can be set with an independent shape order n. i To improve the shape adjustment capability of the transmittance function of the three-channel filter;

[0051] Step 2.2: Establish the correspondence between the transmittance function parameters and the bandwidth parameters;

[0052] To facilitate defining the passband width of the three-channel filter transmittance function in the form of process parameters, the full width at half maximum (FWHMi) of the i-th channel is denoted as FWHMi. The scale parameter... The following relationship exists between the full width at half maximum (FWHMi) and the half height at half maximum (FWHMi):

[0053] ;

[0054] Step 2.3, combining steps 2.1 and 2.2, uses the center wavelength. Peak transmittance Half-height and full-width (FWHMi) and shape order n i Transmittance function of a three-channel filter Provide a unified parameterized description;

[0055] Step 2.4: Establish the process manufacturing constraints for the transmittance function of the three-channel filter;

[0056] To ensure the manufacturability of the transmittance function of the three-channel filter, the following constraints are imposed on the transmittance function of each channel:

[0057] Transmittance range: ;

[0058] Peak transmittance range: ;

[0059] Half-height and full-width range: ;

[0060] Shape order range: ;

[0061] Passband edge slope range: ;

[0062] Out-of-band transmittance integral range: ;

[0063] in, This represents the upper limit of peak transmittance. and These are the lower and upper limits of the half-height and full-width, respectively. and These are the lower and upper limits of the shape order, respectively. Minimum slope constraint for the passband edge. This represents the upper limit of the out-of-band transmittance integration allowed for the i-th channel. Indicates the wavelength position at the edge of the passband of the i-th channel;

[0064] The process can manufacture constraints to limit the physical range of the transmittance function, peak transmittance, passband width range, top and edge shape range, edge transition steepness, and out-of-band transmittance suppression capability, thereby avoiding unmanufacturable transmittance functions with excessively high peak values, excessively narrow or wide passbands, excessively gentle edges, and excessive out-of-band transmittance.

[0065] Step 2.5: Determine the material boundary constraints, system boundary constraints, and channel spacing constraints;

[0066] First, material boundary constraints:

[0067] To suppress the near-infrared extension of the R channel and prevent the transmittance function from entering the boundary region near the effective cutoff wavelength of the detector chain, boundary constraints are applied to the long-wavelength end of the R channel; let the short-wavelength end boundary and the long-wavelength end boundary of the i-th channel be denoted as . and Under symmetric parameterization, it is expressed as:

[0068] ;

[0069] ;

[0070] Then channel R satisfies:

[0071] ;

[0072] in, This is the effective cutoff wavelength of the detector chain using GaAs photocathode as the photosensitive material. To pre-set a safety margin, This represents the long-wavelength end boundary of the R channel;

[0073] Second, system boundary constraints:

[0074] To mitigate the impact of enhanced absorption and scattering on the short-wavelength side of the input window material and coating on the B channel, boundary constraints are applied to the short-wavelength end of the B channel:

[0075] ;

[0076] in, The short-wavelength operating boundary is determined by the system input window and the relevant coating transmission characteristics. To pre-set a safety margin, Boundary constraints for the shortwave end of channel B;

[0077] Third, channel spacing constraints:

[0078] To reduce spectral correlation among the three channels and improve the resolvability of subsequent optimization solutions, a minimum spacing constraint is applied to the center wavelengths of adjacent channels:

[0079] ;

[0080] ;

[0081] ;

[0082] in, The minimum permissible interval between the center wavelengths of adjacent channels; , and These are the center wavelengths of the B channel, G channel, and R channel, respectively.

[0083] By constraining material boundaries, system boundaries, and channel spacing, the transmittance function of the three-channel filter is matched with the effective working range of the detector chain in terms of spectral layout, and adjacent channels are prevented from being too close or too overlapping, thereby improving the stability of the subsequent construction of the three-channel nominal response function and the comprehensive optimization solution.

[0084] Step 2.6: Output the family of three-channel filter transmittance functions that satisfy the constraints;

[0085] The three-channel filter transmittance function family includes the R-channel filter transmittance function. G-channel filter transmittance function and the transmittance function of the B-channel filter ;

[0086] , and All parameters satisfy the parameterization form and constraints described in steps 2.1 to 2.5, and are used as inputs for subsequent construction of the three-channel nominal response function, calculation of evaluation index, and comprehensive optimization solution.

[0087] Furthermore, in S3, the method for constructing the three-channel nominal response function is as follows:

[0088] The three-channel nominal response function refers to the normalized spectral electronic characterization function. With filter transmittance function The theoretical response function obtained by direct multiplication is used to characterize the reception and distribution of effective spectral energy by each channel under the combined effects of the current low-light environment, detector quantum efficiency characteristics and filter transmittance function, and serves as the basis for subsequent evaluation index calculation and comprehensive optimization solution.

[0089] Define the nominal response function of the i-th channel as follows: :

[0090] ;

[0091] The nominal response function of the i-th channel Used to characterize the uniform operating band Within the i-th channel, the relative received energy distribution of the effective spectral energy is obtained; thus, the nominal response functions of the three channels R, G, and B are obtained respectively. , and .

[0092] Furthermore, in S3, the calculation methods for the single-channel relative flux and the total relative flux of the three channels are as follows:

[0093] In a unified working band Within, the relative flux index of the i-th channel is :

[0094] ;

[0095] In a unified working band The total relative flux index of the three channels is as follows: :

[0096] .

[0097] in, Indicates the first The integral representation of the channel nominal response function over the unified operating band is used to characterize the first... The relative effective throughput of the channel; It represents the total relative flux of the three channels, used to characterize the overall available signal level of the three channels;

[0098] The total relative flux of the three channels The larger the value, the higher the effective spectral energy that the system can obtain after being distributed by the three-channel filter under the current environment and detection chain conditions, which helps to reduce the back-end gain requirement to achieve the same signal-to-noise ratio target.

[0099] Furthermore, in S3, the calculation method for the channel energy ratio is as follows:

[0100] To characterize the energy distribution relationship among the three nominal response functions, the energy ratio of the i-th channel is defined as follows: :

[0101] ;

[0102] in, This represents the proportion of the relative flux of the i-th channel in the total relative flux of the three channels;

[0103] R-channel energy ratio G-channel energy ratio Energy ratio of B channel These are used to characterize the energy distribution among the three channels and serve as the basis for subsequently constructing energy balance evaluation indicators; when the preset target energy ratios of the three channels are respectively , and At that time, an energy balance evaluation index is constructed based on the deviation between the energy ratio of each channel and the corresponding target energy ratio, so as to constrain the energy distribution of each channel determined by the nominal response function of the three channels from being excessively unbalanced;

[0104] The preset three-channel target energy ratio , and All settings are pre-defined based on the detector's quantum efficiency characteristics, system mission requirements, and subsequent color recovery requirements.

[0105] Furthermore, in S3, the calculation method for the channel correlation evaluation quantity is as follows:

[0106] To characterize the pairwise correlation between the nominal response functions of different channels, in a unified operating band The weighted correlation coefficient between the i-th channel and the j-th channel is defined internally. :

[0107] ;

[0108] in, The weighting function represents the weight allocation of different wavelength positions in the correlation calculation;

[0109] like If a constant of 1 is taken, then the positions of each wavelength within the unified working band are calculated with equal weights;

[0110] like By weighting the parameters based on the importance of key spectral bands, the quantum efficiency characteristics of the detector, or the system mission requirements, the constraint effect of correlation evaluation on key bands is enhanced.

[0111] The weighted correlation coefficient This is a channel correlation evaluation metric, representing the pairwise similarity between the nominal response functions of channel i and channel j. The larger the value, the stronger the correlation between the nominal response functions of the two channels and the more obvious the spectral overlap; The smaller the value, the higher the degree of separation between the two channels;

[0112] Furthermore, in S3, the calculation method for the overall separability evaluation quantity of the three channels is as follows:

[0113] To characterize the joint separability of the overall three-channel nominal response functions, a weighted Gram matrix G is constructed based on the weighted correlation coefficients between the channels. The elements of matrix G are defined as G... ij :

[0114] ;

[0115] Where, when i=j, G ij =1 represents the diagonal element, indicating the normalized correlation between each channel and itself; Off-diagonal elements represent the weighted correlation between different channels;

[0116] The condition number of the eigenvalues ​​of the weighted Gram matrix G for:

[0117] ;

[0118] in, and Let G and G represent the largest and smallest eigenvalues ​​of the weighted Gram matrix G, respectively.

[0119] The condition number It is a three-channel overall separability evaluation quantity, used to characterize the overall joint separability of the three-channel nominal response functions and the numerical stability of the subsequent solution process; The larger the value, the stronger the overall correlation and the worse the separability among the three-channel nominal response functions, and the more sensitive the subsequent optimization solution is to noise and parameter perturbations. The smaller the value, the better the overall separability of the three-channel nominal response function, and the higher the stability of subsequent solutions.

[0120] In the subsequent comprehensive optimization process, with The weighted correlation coefficient serves as the primary indicator for evaluating channel separation. As an auxiliary evaluation metric for pairwise channel correlation analysis, it is used to explain the local overlap between different channels or as a basis for additional constraints, thereby avoiding the repeated characterization of channel overlap in the comprehensive optimization objective.

[0121] Furthermore, in S4, the process of constructing the comprehensive optimization objective function is as follows:

[0122] Step 4.1: Construct channel energy balance evaluation indicators :

[0123] To constrain the energy distribution of each channel, determined by the nominal response function of the three channels, from becoming excessively unbalanced, the energy ratio of the three channels obtained in S3 is used as a basis. Constructing channel energy balance evaluation indicators :

[0124] ;

[0125] The channel energy balance evaluation index Used to characterize the degree of deviation between the channel energy allocation corresponding to the current three-channel nominal response function and the preset target energy ratio; The smaller the value, the closer the energy distribution of the three channels is to the preset target ratio, and the lower the degree of imbalance in the response of the three channels during subsequent color restoration;

[0126] Step 4.2: Construct an overall three-channel separability evaluation index :

[0127] To characterize the overall joint separability of the three-channel nominal response functions, the overall three-channel separability evaluation metric obtained in S3 is used. As an evaluation index for the overall separability of the three channels ;

[0128] ;

[0129] The larger the value, the stronger the overall correlation and the worse the joint separability among the three-channel nominal response functions; The smaller the value, the better the overall separability of the three-channel nominal response function, and the higher the stability of subsequent solutions;

[0130] The channel correlation evaluation value obtained from S3 It is used as an auxiliary analytical measure or additional constraint for the local correlation between different channels to explain the local overlap between adjacent channels, and is no longer used as a basis for analysis of the relationship between channels. Parallel main optimization indices with equal weights are used to avoid redundant representation of channel overlap in the comprehensive optimization objective function;

[0131] Step 4.3: Construct the curve morphology of the transmittance function and the evaluation index of process feasibility. :

[0132] To further improve the fabrication feasibility of the filter transmittance function while satisfying the manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints given in S2, a curve morphology of the transmittance function and a fabrication feasibility evaluation index are constructed based on the edge transition characteristics and out-of-band transmission suppression characteristics of the transmittance function. ;

[0133] ;

[0134] in, and They are respectively and Sub-item weight coefficients For edge shape penalty term, This is a penalty item for bypassing the zone;

[0135] The edge shape penalty item This term is used to characterize the deviation between the transition characteristics of the transmittance function of each channel filter at the passband edge and the edge morphology achievable by a preset process; the out-of-band transmittance penalty term. Used to characterize the residual transmittance of each channel filter outside the nominal passband;

[0136] By constructing the curve morphology of the transmittance function and evaluating the feasibility of the process. Under the premise of meeting the hard constraints set by S2, further suppress filter transmittance functions with excessively gentle edges, excessively large out-of-band residual transmittance, or curve morphology that are not conducive to manufacturing realization, thereby improving the process feasibility of the optimization results.

[0137] Step 4.4, construct the comprehensive optimization objective function:

[0138] Step 4.4.1, Define the parameter vector to be optimized. :

[0139] ;

[0140] Step 4.4.2, in the parameter vector to be optimized Above, construct a comprehensive optimization objective function. :

[0141] ;

[0142] in, , , and These are the total relative fluxes of the three channels. Channel energy balance evaluation indicators Three-channel overall separability evaluation index The curve morphology of the transmittance function and the evaluation index of process feasibility The weighting coefficients are used to weigh the total relative flux of the three channels, channel energy balance, overall separability of the three channels, and the curve shape of the transmittance function against the process feasibility according to the task requirements.

[0143] The weighting coefficient , , and The parameters are set according to the total throughput priority, channel separation requirements, energy balance requirements and process implementation requirements, and are adjusted based on experimental experience, numerical simulation or verification results;

[0144] The comprehensive optimization objective function Constructed in a minimal form, where, The form of a negative sign is used to enter the objective function, which is used to drive the maximization of the total relative flux of the three channels within the minimization framework; , as well as The penalty term is incorporated into the comprehensive optimization objective function. This is to constrain the energy distribution relationship corresponding to the nominal response function of the three channels, the overall separability, and the rationality of the curve shape of the filter transmittance function and the feasibility of the process.

[0145] Furthermore, in S4, the solution process for the comprehensive optimization objective function is as follows:

[0146] The process of solving the comprehensive optimization objective function involves jointly solving the transmittance function parameters of the three-channel filter under the constraints of process manufacturability, material boundary, system boundary, and channel spacing given in S2, in order to obtain the optimized three-channel filter transmittance function and its corresponding optimized three-channel nominal response function that meet multiple performance index requirements.

[0147] S4.1, Perform a global search under constraints to obtain a candidate parameter vector as initial values:

[0148] Within the manufacturable design domain, a global search is performed using a genetic algorithm or a differential evolution algorithm to obtain a set of candidate parameters that satisfy the constraints and the corresponding candidate comprehensive optimization objective function values.

[0149] The global search is used to search for candidate parameter vectors that satisfy the constraints in the parameter space, reducing the sensitivity of the solution process to the selection of initial values, and providing initial parameters for subsequent local fine-tuning.

[0150] S4.2, perform local fine-tuning of the comprehensive optimization objective function:

[0151] While maintaining the manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints given in S2, the comprehensive optimization objective function is locally fine-tuned to obtain the optimal parameter vector. Optimized three-channel filter transmittance function and the optimized three-channel nominal response function ;

[0152] The local fine-tuning employs a local optimization method based on sequential quadratic programming or trust region-based constraint optimization to further improve the accuracy of parameter solving and enable the comprehensive optimization objective function to converge to the optimal solution within the current constraint domain.

[0153] By combining a two-stage solution strategy of global search and local fine-tuning, the risk of getting trapped in local optima can be reduced while improving the stability of the final parameter solution and the manufacturability of the optimization results.

[0154] The optimal parameter vector Optimized three-channel filter transmittance function And the optimized three-channel nominal response function It serves as the input for physical realizability correction, channel energy compensation, and final transmittance data shaping in S5.

[0155] Furthermore, in S5, the final qualitative process of the transmittance function of the three-channel filter is as follows:

[0156] While maintaining the overall shape and separation performance of the S4 optimization results, further meet the requirements of membrane system realization constraints, batch production consistency requirements and B channel energy compensation requirements to obtain the final RGB filter transmittance data;

[0157] Peak limiting calibration is first performed on the R and G channels to meet the upper limit of peak performance and batch consistency requirements of the film system; for the B channel, which is susceptible to insufficient energy due to the influence of the short-wavelength system boundary, energy state discrimination is then performed, and limiting compensation is implemented when the constraint conditions are met.

[0158] Step 5.1, perform peak-limited calibration on the R and G channels:

[0159] To optimize the transmittance function of the R-channel filter output by S4 and the transmittance function of the G-channel filter To more closely resemble the actual membrane preparation level and improve the consistency of formulation in mass production, peak limiting calibration is performed on the R channel and G channel respectively;

[0160] In the final design stage, the The value was set to 0.95 to ensure that the final product better reflects the actual membrane preparation level.

[0161] Define the transmittance function of the calibrated R-channel filter respectively. and the transmittance function of the G-channel filter :

[0162] ;

[0163] ;

[0164] in, and They are respectively and The scaling factor satisfies:

[0165] ;

[0166] ;

[0167] This guarantees:

[0168] ;

[0169] ;

[0170] Among them, the upper limit of peak transmittance is That is, the one given in S2 ;

[0171] The peak-limited calibration does not change the overall shape of the transmittance functions of the R-channel and G-channel filters; it only applies a limited scaling to the peak amplitude to ensure that the peak values ​​of both channels meet the upper limit of peak transmittance. This meets the requirements and facilitates consistency control in subsequent batch production;

[0172] Step 5.2, Determining the energy state of channel B:

[0173] Considering that the B channel is more susceptible to the influence of the short-wavelength system boundary, resulting in lower effective spectral energy, to determine whether the optimized B channel filter transmittance function has an energy deficiency problem, the optimized three-channel nominal response function output by S4 was used. Define the optimized channel energy of the i-th channel as: :

[0174] ;

[0175] Right now:

[0176] ;

[0177] The optimized channel energy Used to characterize the channels in a unified operating band before finalization. The relatively effective spectral energy available within;

[0178] For channel B, if the channel energy E is optimized... B After optimization of the G channel, the channel energy E G satisfy:

[0179] ;

[0180] Then it is determined that channel B has insufficient energy; among which, A preset energy ratio threshold is used to characterize the minimum allowable energy ratio of the B channel relative to the G channel, and its value range is [value range missing]. ;

[0181] The preset threshold, preset upper limit, and preset safety margin are set according to the detector characteristics, film preparation capability, system task requirements, and experimental verification results.

[0182] Step 5.3: Perform limiting compensation processing on channel B to obtain the qualitative transmittance function of channel B filter;

[0183] When it is determined that the B channel has insufficient energy, and the B channel compensation does not cause the overall separability evaluation value of the three channels to exceed the preset threshold, the transmittance function of the optimized B channel filter is... After performing limited compensation processing, the transmittance function of the B-channel filter after finalization is obtained. ;

[0184] The restricted compensation process is performed within the constraints given in S2 and satisfies the overall separation performance requirements formed in S4.

[0185] The restricted compensation process improves the effective spectral energy of the B channel by restrictively adjusting the local transmittance distribution and edge transition shape within the effective passband of the B channel, while controlling the increase in the additional overlap between the B channel and the G channel.

[0186] The restricted compensation process simultaneously satisfies the following conditions:

[0187] First, it does not disrupt the preset spectral boundaries of channel B at the shortwave and longwave ends;

[0188] Second, do not exceed the peak transmittance limit. and minimum slope constraint at the edge;

[0189] Third, prevent the out-of-band transmittance integral value from exceeding the preset upper limit;

[0190] Fourth, the overall separability evaluation value of the three channels should not exceed the preset threshold;

[0191] If the B-channel energy deficiency judgment is not triggered, the transmittance function of the finalized B-channel filter will be... :

[0192] ;

[0193] If the insufficient energy determination of channel B is triggered, then under the premise of satisfying the constraints, ... After local compensation adjustments, the transmittance function of the B-channel filter after finalization is obtained. ;

[0194] The local compensation process is applied to the high-transmittance region and its adjacent transition region within the effective passband of channel B, in order to increase the effective spectral energy of channel B without significantly increasing the degree of overlap with channel G.

[0195] Step 5.4, output the final RGB filter transmittance data:

[0196] After the peak limiting calibration of the R and G channels in step 5.1 and the limiting compensation processing of the B channel in step 5.3, the final RGB filter transmittance data is obtained. ,in ;

[0197] Final R-channel filter transmittance function and the transmittance function of the G-channel filter Obtained from the peak-limited calibration in step 5.1;

[0198] Final B-channel filter transmittance function It is obtained from the B-channel limitation compensation process in step 5.3.

[0199] The final RGB filter transmittance data meets the requirements of film system realization constraints, batch consistency requirements, and channel energy configuration requirements, and can be directly used for subsequent filter fabrication, array layout design, and imaging system integration.

[0200] An RGB filter design device for low-light night vision includes a data processing unit, a response calculation unit, an evaluation unit, an objective function unit, a first optimization unit, a second optimization unit, and an output unit. Each unit can be implemented by a processor, a memory, and a program module stored in the memory. The output of the data processing unit is connected to the input of the response calculation unit, the output of the response calculation unit is connected to the input of the evaluation unit, the output of the evaluation unit is connected to the input of the objective function unit, the output of the objective function unit is connected to both the first optimization unit and the second optimization unit, and the output of the second optimization unit is connected to the output unit and fed back to the response calculation unit to form an iterative solution closed loop.

[0201] The data processing unit is used to acquire and preprocess the environmental spectrum and the end-to-end quantum efficiency of the GaAs photocathode, and calculate and normalize the spectral electron flux curve S(λ) on a unified wavelength grid.

[0202] The response calculation unit is used to calculate the transmittance based on the current three-channel parameterized transmittance function. Calculate the effective response hi(λ) for each channel;

[0203] Evaluation unit, used to evaluate the nominal response hi(λ) and parameterized transmittance function Extract the evaluation metrics required for optimization; including those used in parallel:

[0204] The first evaluation subunit is used to calculate the throughput PUi of each channel and the total throughput PUsum;

[0205] The second evaluation subunit is used to calculate the flux ratio / balance index Jbal, which is a measure of the deviation of PUi / PUj from the target ratio.

[0206] The third evaluation subunit is used to calculate the channel correlation / overlap index Joverlap, such as the weighted correlation coefficient ρij or the weighted Gram matrix condition number κ2(G);

[0207] The fourth evaluation subunit is used to calculate the transmittance curve shape constraint index Jshape, which includes edge slope constraint, out-of-band leakage integral upper limit constraint, and smoothness constraint metric.

[0208] The objective function unit is used to assemble the above indicators into a comprehensive optimization objective function according to the set weights;

[0209] The first optimization unit is used to perform a global search on the comprehensive optimization objective function within the manufacturable design domain to obtain the global near-optimal solution and the initial form of transmittance for the parameter vector θ.

[0210] The second optimization unit performs local fine-tuning based on the output of the first optimization unit, and applies upper limits constraints on peak transmittance, minimum edge slope, and out-of-band leakage integral, thereby making the parameterized transmittance function... The curve converges to a shape that can be manufactured.

[0211] The output unit is used to output the final three-channel transmittance Ti(λ), nominal response hi(λ), and corresponding index results; and to send the updated parameter vector θ back to the iterative update path for the next round of solution.

[0212] The above "each unit" is a functional description and can be implemented by dedicated hardware, programmable hardware, or software modules running on a processor and their combinations; the numbers in the figure are only for distinguishing functions and do not limit the specific implementation form.

[0213] Compared with the prior art, the technical effects of the present invention are as follows:

[0214] This application can improve photon utilization efficiency and control channel energy ratio, reduce electronic gain or integration time under the same imaging quality; reduce channel crosstalk and improve unmixing stability through correlation / overlap measurement and morphology constraint; combine GaAs photocathode characteristics to implement R-channel upper edge control and B-channel enhancement to ensure the three channels are balanced and effective.

[0215] The method and apparatus of this invention inherently introduce manufacturability constraints such as peak value upper limit, bandwidth range, out-of-band leakage integral upper limit, and edge slope during the solution process. The results can be directly used for membrane system design and testing. Simultaneously, it outputs objective quantitative indicators such as flux, proportion, and correlation, facilitating engineering verification and subsequent consistency comparison. Overall, the solution offers advantages such as flexible implementation, configurable parameters, and scene adaptability, significantly improving the engineering accessibility and stability of low-light color imaging systems. Attached Figure Description

[0216] Figure 1 This is a schematic diagram of the filter design process provided in the embodiments of this application;

[0217] Figure 2 This is a schematic diagram of the functional structure of the filter design device provided in the embodiments of this application;

[0218] Figure 3 A schematic diagram of the normalized spectral electronic characterization function provided in the embodiments of this application;

[0219] Figure 4 This is a schematic diagram of the transmittance function of a three-channel filter provided in an embodiment of this application;

[0220] Figure 5 This is a schematic diagram of the final RGB filter transmittance data provided in the embodiments of this application;

[0221] Figure 6 This is a schematic diagram of the three-channel nominal response function provided in an embodiment of this application;

[0222] Figure 7 This is a schematic diagram of channel correlation evaluation provided in an embodiment of this application;

[0223] Figure 8 This is a schematic diagram of the channel energy ratio provided in the embodiments of this application;

[0224] Figure 9 This is a schematic diagram of the transmittance function parameters of a three-channel filter provided in an embodiment of this application. Detailed Implementation

[0225] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0226] The following embodiments are only used to illustrate the technical solutions and implementation methods of the present invention, and are intended to aid understanding, and do not constitute a limitation on the claims. Without changing the technical essence and effects, those skilled in the art can make equivalent substitutions or adjustments to the step order, parameter values, module division, and combination methods in the embodiments, all of which should be covered within the protection scope of the present invention.

[0227] The numbering of "first", "second", etc. in the text is only used to distinguish objects of the same kind and does not indicate a limitation on order or quantity; "including / contains" and its variations indicate non-exclusivity; interval descriptions (such as A to B) include the endpoints by default; when expressions such as "about" or "approximately" appear, variations are allowed within the range of measurement / process tolerances recognized in the field.

[0228] The steps in the flowchart can be executed sequentially, in parallel, or combined / split during implementation; the objective function, weights, thresholds, bandwidth, peak transmittance, edge slope, and correlation coefficient are configurable parameters, and their specific values ​​can be determined based on the application scenario and manufacturing capabilities unless otherwise explicitly specified.

[0229] The devices or module units described in this specification can be implemented separately or integrated in physical hardware form, or deployed in a distributed manner. Coupling or connection between different units can be achieved through electrical connections, mechanical connections, communication connections, or indirect connections via intermediate interfaces. Any method that enables the corresponding function should be considered within the scope of this invention.

[0230] Modules described as functional units can be implemented in hardware, software, or a combination of both. If implemented as software modules, they can be stored in computer-readable storage media and loaded and executed by a processor to perform their respective functions.

[0231] The embodiments are illustrated using an EBCMOS system with a built-in GaAs photocathode as an example to demonstrate its high response background in the short-wave near-infrared region. For systems using other photocathode materials, the same technical effect can be obtained by referring to the design process and constraints of this invention for parameter mapping and adjustment.

[0232] Figure 2 The second aspect of this application provides an RGB filter design device for low-light night vision, comprising: a data processing unit 401, a transmittance function construction unit 402, a response calculation unit 403, an evaluation quantity calculation unit 404, a first optimization unit 405, a second optimization unit 406, a first processing unit 407, a second processing unit 408, a third processing unit 409, and an output unit 410; each unit may be implemented by a processor, a memory, and a program module stored in the memory. The output of the data processing unit 401 is connected to the input of the transmittance function construction unit 402 and the response calculation unit 403. The output of the transmittance function construction unit 402 is connected to the input of the response calculation unit 403. The output of the response calculation unit 403 is connected to the input of the evaluation quantity calculation unit 404. The output of the evaluation quantity calculation unit 404 is connected to the input of the first optimization unit 405. The output of the first optimization unit 405 is connected to the input of the second optimization unit 406. The output of the second optimization unit 406 is connected to the input of the first processing unit 407. The output of the first processing unit 407 is connected to the input of the second processing unit 408. The output of the second processing unit 408 is connected to the input of the third processing unit 409. The output of the third processing unit 409 is connected to the output unit 410 and can be fed back to the response calculation unit 403 to form an iterative solution closed loop.

[0233] The data processing unit 401 is used to acquire environmental spectral data and detector quantum efficiency functions, perform peak normalization processing at a unified wavelength sampling point, and construct a normalized spectral electronic characterization function by combining a metric domain conversion factor. The environmental spectral data is the environmental spectral distribution under nighttime light environment or darkroom environment, and the detector quantum efficiency function is the quantum efficiency function of an electron-bombarded CMOS detector chain using gallium arsenide (GaAs) photocathode as the photosensitive material at different wavelengths.

[0234] The transmittance function construction unit 402 is used to establish a parameterized expression of the transmittance function of the three-channel filter and, in conjunction with constraints, construct a family of manufacturable three-channel filter transmittance functions. These constraints include process manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints. The R-channel filter transmittance function, G-channel filter transmittance function, and B-channel filter transmittance function in the family of three-channel filter transmittance functions serve as inputs for subsequent nominal response function construction, evaluation quantity calculation, and comprehensive optimization solution.

[0235] The response calculation unit 403 is used to multiply the normalized spectral electronic characterization function with the three-channel filter transmittance function to construct the three-channel nominal response function. The three-channel nominal response function is a theoretical response function obtained by directly multiplying the normalized spectral electronic characterization function with the filter transmittance function. It is used to characterize the reception and distribution of effective spectral energy by each channel under the combined effects of the current low-light environment, detector quantum efficiency characteristics, and filter transmittance function, and serves as the basis for subsequent evaluation quantity calculations and comprehensive optimization solutions.

[0236] The evaluation quantity calculation unit 404 is used to calculate the single-channel relative flux, the total relative flux of the three channels, the channel energy ratio, the channel correlation evaluation quantity, and the overall separability evaluation quantity of the three channels based on the nominal response function of the three channels. The evaluation quantity calculation unit 404 includes a first calculation subunit 404-1, a second calculation subunit 404-2, a third calculation subunit 404-3, and a fourth calculation subunit 404-4; wherein, the first calculation subunit 404-1 is used to calculate the single-channel relative flux and the total relative flux of the three channels, the second calculation subunit 404-2 is used to calculate the channel energy ratio, the third calculation subunit 404-3 is used to calculate the channel correlation evaluation quantity, and the fourth calculation subunit 404-4 is used to calculate the overall separability evaluation quantity of the three channels.

[0237] The first optimization unit 405 is used to construct a comprehensive optimization objective function and, under the constraints established by the transmittance function construction unit 402, to perform a global search on the transmittance function parameters of the three-channel filter to obtain candidate parameter results that satisfy the constraints. The comprehensive optimization objective function is used to jointly weigh the total relative flux of the three channels, the channel energy balance evaluation index, the overall separability evaluation index of the three channels, and the curve shape and process feasibility evaluation index of the transmittance function.

[0238] The second optimization unit 406, based on the output of the first optimization unit 405, performs local fine-tuning of the comprehensive optimization objective function while maintaining process manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints, to obtain the optimized three-channel filter transmittance function and its corresponding optimized three-channel nominal response function. The first optimization unit 405 and the second optimization unit 406 correspond to the global search stage and the local fine-tuning stage in the comprehensive optimization solution process, respectively.

[0239] The first processing unit 407 is used to perform peak limiting calibration on the optimized R-channel filter transmittance function and the optimized G-channel filter transmittance function output by the second optimization unit 406, so as to obtain the calibrated R-channel filter transmittance function and the calibrated G-channel filter transmittance function, and ensure that neither of them exceeds the preset peak transmittance limit.

[0240] The second processing unit 408 is used to determine the energy state of channel B based on the optimized nominal response function of the three channels output by the second optimization unit 406; when the optimized channel energy of channel B is lower than the optimized channel energy of channel G by a preset energy ratio threshold, it is determined that channel B has insufficient energy.

[0241] The third processing unit 409 is used to perform restricted compensation processing on the optimized B-channel filter transmittance function under the condition that the B-channel has insufficient energy and the compensation processing does not cause the overall separability evaluation value of the three channels to exceed a preset threshold, so as to obtain the finalized B-channel filter transmittance function. The restricted compensation processing simultaneously satisfies the following conditions: it does not destroy the preset spectral boundary of the B-channel, does not exceed the upper limit of peak transmittance and the minimum slope constraint at the edge, does not cause the out-of-band transmittance integral value to exceed the preset upper limit, and does not cause the overall separability evaluation value of the three channels to exceed the preset threshold. If the B-channel energy deficiency determination is not triggered, the optimized B-channel filter transmittance function is kept as the finalized B-channel filter transmittance function.

[0242] Output unit 410 is used to output the final RGB filter transmittance data. The final RGB filter transmittance data includes the final R channel filter transmittance function, the final G channel filter transmittance function, and the final B channel filter transmittance function; wherein, the final R channel filter transmittance function and the final G channel filter transmittance function are output by the first processing unit 407, and the final B channel filter transmittance function is output by the third processing unit 409.

[0243] Figure 3This diagram illustrates the normalized spectral electronic characterization function provided in this application embodiment. In the detector chain using GaAs photocathode as the material, the normalized spectral electronic characterization in the short-wavelength near-infrared region accounts for a relatively high proportion. Based on this distribution characteristic, the transmittance function of the three-channel filter can more fully utilize the effective energy in the short-wavelength near-infrared region during subsequent construction and optimization, and further reflect this... Figure 5 and Figure 6 The transmittance function configuration and nominal response function distribution shown are ultimately reflected in the channel energy ratio results.

[0244] Figure 4 This diagram illustrates the transmittance function of a three-channel filter provided in this embodiment. The three channels are parameterized using a family of manufacturable bandpass curves, and a family of transmittance functions for the three-channel filter is constructed in conjunction with constraints. The result shows that the three channels are sequentially distributed at their center positions, with resolvable intervals between adjacent channels, satisfying manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints. This diagram is used to illustrate the spectral characteristics and constraint satisfaction of the three-channel filter transmittance function during the optimization stage, and does not limit the specific values.

[0245] Figure 5 This is a schematic diagram of the final RGB filter transmittance data provided in the embodiments of this application. After comprehensive optimization and finalization, the transmittance function of the three-channel filter maintains a reasonable central distribution relationship and spectral band configuration relationship, with clear separation between adjacent channels. This can improve the total system throughput and channel energy ratio while meeting manufacturing constraints, providing a basis for subsequent nominal response function construction and energy configuration.

[0246] Figure 6 This is a schematic diagram of the three-channel nominal response function provided in an embodiment of this application. Under the action of the normalized spectral electronic characterization function, the three-channel nominal response function reflects the reception and distribution relationship of each channel to the effective spectral energy; adjacent channels maintain resolvable separation within the response domain, which is beneficial to reduce spectral overlap between channels and improve the stability of the subsequent color recovery process.

[0247] Figure 7 This diagram illustrates the channel correlation evaluation provided in this application embodiment. Each element in the diagram represents the correlation evaluation result calculated under the weighting function between each channel. Diagonal elements represent the correspondence between each channel and itself, while off-diagonal elements correspond to the degree of correlation between channels. As can be seen from the diagram, the correlation evaluation results between channels meet the preset constraints, indicating that while maintaining the necessary spectral band connections, the three channels do not exhibit excessive overlap that would affect the stability of subsequent separation.

[0248] Figure 8This diagram illustrates the channel energy distribution in an embodiment of this application. ER, EG, and EB represent the relative flux results for channels R, G, and B, respectively. Under the premise of ensuring channel resolvability and controlled correlation, the channel energy distribution is consistent with the distribution characteristics of the normalized spectral electronic characterization function, reflecting a comprehensive balance between the total system flux, channel energy distribution, and channel separability.

[0249] Figure 9 This diagram illustrates the transmittance function parameters of a three-channel filter provided in an embodiment of this application. The diagram shows the spectral band parameter ranges corresponding to the R, G, and B channels, as well as explanations of parameter indices related to channel energy allocation, used to characterize the parameter configuration results of the transmittance function for each channel.

[0250] A third aspect of this application provides a computer-readable storage medium storing computer program instructions that, when loaded and executed by a processor, enable it to perform the steps of the aforementioned RGB filter design method. The computer-readable storage medium may be a disk, optical disk, read-only memory, random access memory, solid-state drive, or other medium capable of storing program code. By executing the program instructions stored in the storage medium on a computer or server, the steps of the method of this invention can be implemented.

[0251] In a representative implementation of this invention, the optimization weights and constraint parameters can be set according to specific application scenarios, manufacturing capabilities, and system performance requirements. These parameters include weight coefficients in the comprehensive optimization objective function, the energy state discrimination threshold for the B channel, the upper limit of peak transmittance, the upper limit of out-of-band transmittance integral, and the minimum edge slope, etc. Each parameter can be adjusted while satisfying process manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints.

[0252] Those skilled in the art can make equivalent adjustments or substitutions to the aforementioned methods, processes, and apparatus according to specific application scenarios, and all such adjustments should be considered to fall within the protection scope of this invention. Furthermore, the terms "may" and "optional" mentioned in this specification are intended to indicate that the feature is optional rather than essential, and do not constitute a limitation on the technical solution of this invention.

[0253] The directional and relative terms (such as "up", "down", "inner", "outer") used in the instruction manual are used only for ease of description and do not indicate or imply that the corresponding components must have a specific orientation, nor should they be interpreted as a limitation on the structure.

[0254] Provided there is no contradiction, the technical features of the various embodiments in this specification can be combined arbitrarily to form new embodiments, all of which fall within the protection scope of this invention.

[0255] The above are merely preferred embodiments of the present invention. For those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and such improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for designing RGB filters for low-light night vision, characterized in that, Includes the following steps: S1, acquire environmental spectral data and detector quantum efficiency function, perform peak normalization processing at a unified wavelength sampling point, and construct a normalized spectral electronic characterization function by combining the metric domain conversion factor; S2, establish a parameterized expression for the transmittance function of a three-channel filter, and construct a family of manufacturable three-channel filter transmittance functions in combination with constraints; the constraints include process manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints; S3. Multiply the normalized spectral electronic characterization function with the transmittance function of the three-channel filter to construct the nominal response function of the three channels, and calculate the single-channel relative flux, the total relative flux of the three channels, the channel energy ratio, the channel correlation evaluation quantity and the overall separability evaluation quantity of the three channels. S4. Construct a comprehensive optimization objective function; under the constraints obtained in S2, solve for the comprehensive optimization of the transmittance function parameters of the three-channel filter. S5 sequentially performs peak limiting calibration, B-channel energy discrimination, and limiting compensation processing on the optimized three-channel filter transmittance function output from S4 to obtain the final RGB filter transmittance data that meets the requirements of film system realization and batch consistency.

2. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S1, the process of constructing the normalized spectral electronic characterization function includes: The environmental spectral data refers to the environmental spectral distribution under nighttime light conditions or darkroom conditions. ; The quantum efficiency function of the detector is the quantum efficiency function of the electron-bombarded CMOS detector chain using gallium arsenide (GaAs) photocathode as the photosensitive material at different wavelengths. ; The environmental spectral distribution With quantum efficiency function All in wavelength As the independent variable, it is used to characterize the relationship between the external radiation input and the quantum efficiency characteristics of the detector under low light conditions in the wavelength dimension; Step 1.1, analyze the environmental spectral distribution. With quantum efficiency function Mapping to a unified wavelength sampling point and defining a unified operating band. ; The mapping is the mapping of the environmental spectral distribution. With quantum efficiency function In a unified working band Resampling is performed within the process; Step 1.1.1, analyze the environmental spectral distribution. With quantum efficiency function The original discrete data is subjected to interpolation and noise reduction processing; Step 1.1.2: Select the band according to the preset unified working band; unified working band Determined by the spectral range of the detector's quantum efficiency; Step 1.2, map the environmental spectral distribution. and quantum efficiency function Peak value normalization was performed separately: Based on environmental spectral distribution In a unified working band Maximum value and quantum efficiency function In a unified working band The maximum value within the range is used as the normalization benchmark for the mapped environmental spectral distribution. and quantum efficiency function Amplitude scaling is performed to obtain the environmental spectral distribution after peak normalization. and the quantum efficiency function after peak normalization : ； ； Where λ is the wavelength. and The maximum value of all of them is 1; Step 1.3: Construct an initial spectral electronic characterization function at a uniform wavelength sampling point to obtain a normalized spectral electronic characterization function; Step 1.3.1: Under the selected metric domain, normalize the environmental spectral distribution of the peak values. Quantum efficiency function after peak normalization and metric domain conversion factor Multiplication yields the initial spectral electronic characterization function. : ； When using power domain representation ; When using photon number field representation ; Where h is Planck's constant and c is the speed of light in a vacuum; Step 1.3.2, characterize the initial spectral electronic function. After peak normalization, the normalized spectral electronic characterization function is obtained. : ； in, satisfy: .

3. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S2, the process of constructing the transmittance function of the three-channel filter is as follows: The three-channel filter includes an R channel, a G channel, and a B channel, with channel indices denoted as follows: ; The R, G, and B are only used to distinguish the three filter channel identifiers and do not limit their passbands to correspond to the red, green, and blue bands in the standard visual sense. The transmittance function of the i-th channel filter is denoted as: Characterizes the wavelength of the i-th channel filter. The transmittance at a given point as a function of wavelength; Step 2.1: Select the parameterized bandpass function family of the three-channel filter transmittance function; Transmittance function of a three-channel filter Parametric characterization is performed using symmetric superGaussian bandpass functions: ； in, The center wavelength of the i-th channel is Let be the peak transmittance of the i-th channel. n is the scale parameter. i Shape order; Step 2.2: Establish the correspondence between the transmittance function parameters and the bandwidth parameters; Let FWHMi denote the full width at half maximum (FWH) of the i-th channel, then the scale parameter... The following relationship exists between the full width at half maximum (FWHMi) and the half height at half maximum (FWHMi): ； Step 2.3, combining steps 2.1 and 2.2, uses the center wavelength. Peak transmittance Half-height and full-width (FWHMi) and shape order n i Transmittance function of a three-channel filter Provide a unified parameterized description; Step 2.4: Establish the process manufacturing constraints for the transmittance function of the three-channel filter; The following constraints are applied to the transmittance function of each channel: Transmittance range: ; Peak transmittance range: ; Half-height and full-width range: ; Shape order range: ; Passband edge slope range: ; Out-of-band transmittance integral range: ; in, This represents the upper limit of peak transmittance. and These are the lower and upper limits of the half-height and full-width, respectively. and These are the lower and upper limits of the shape order, respectively. Minimum slope constraint for the passband edge. This represents the upper limit of the out-of-band transmittance integration allowed for the i-th channel. Indicates the wavelength position at the edge of the passband of the i-th channel; Step 2.5: Determine the material boundary constraints, system boundary constraints, and channel spacing constraints; First, material boundary constraints: Apply boundary constraints to the long-wavelength end of channel R; let the short-wavelength end boundary and long-wavelength end boundary of channel i be denoted as . and Under symmetric parameterization, it is expressed as: ； ； Then channel R satisfies: ； in, This is the effective cutoff wavelength of the detector chain using GaAs photocathode as the photosensitive material. To pre-set a safety margin, This represents the long-wavelength end boundary of the R channel; Second, system boundary constraints: Apply boundary constraints to the shortwave end of channel B: ； in, The short-wavelength operating boundary is determined by the system input window and the relevant coating transmission characteristics. To pre-set a safety margin, Boundary constraints for the shortwave end of channel B; Third, channel spacing constraints: Apply a minimum spacing constraint to the center wavelengths of adjacent channels: ； ； ； in, The minimum permissible interval between the center wavelengths of adjacent channels; , and These are the center wavelengths of the B channel, G channel, and R channel, respectively. Step 2.6: Output the family of three-channel filter transmittance functions that satisfy the constraints; The three-channel filter transmittance function family includes the R-channel filter transmittance function. G-channel filter transmittance function and the transmittance function of the B-channel filter ; , and All satisfy the parameterization form and constraints described in steps 2.1 to 2.

5.

4. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S3, the method for constructing the three-channel nominal response function is as follows: Define the nominal response function of the i-th channel as follows: : ； The nominal response function of the i-th channel Used to characterize the uniform operating band Within the i-th channel, the relative received energy distribution of the effective spectral energy is obtained; thus, the nominal response functions of the three channels R, G, and B are obtained respectively. , and .

5. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S3, the calculation methods for the single-channel relative flux and the total relative flux of the three channels are as follows: In a unified working band Within, the relative flux index of the i-th channel is : ； In a unified working band The total relative flux index of the three channels is as follows: : ； In S3, the calculation method for the channel energy ratio is as follows: Define the energy ratio of the i-th channel as follows: : ； in, This represents the proportion of the relative flux of the i-th channel in the total relative flux of the three channels; R-channel energy ratio G-channel energy ratio Energy ratio of B channel These are used to characterize the energy distribution among the three channels; when the preset target energy ratios of the three channels are respectively , and At that time, an energy balance evaluation index is constructed based on the deviation between the energy ratio of each channel and the corresponding target energy ratio, so as to constrain the energy distribution of each channel determined by the nominal response function of the three channels from being excessively unbalanced; The preset three-channel target energy ratio , and All settings are pre-defined based on the detector's quantum efficiency characteristics, system mission requirements, and subsequent color recovery requirements.

6. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S3, the calculation method for the channel correlation evaluation metric is as follows: In a unified working band The weighted correlation coefficient between the i-th channel and the j-th channel is defined internally. : ； in, The weighting function represents the weight allocation of different wavelength positions in the correlation calculation; like If a constant of 1 is taken, then the positions of each wavelength within the unified working band are calculated with equal weights; like By weighting the parameters based on the importance of key spectral bands, the quantum efficiency characteristics of the detector, or the system mission requirements, the constraint effect of correlation evaluation on key bands is enhanced. The weighted correlation coefficient This is a channel correlation evaluation metric, representing the pairwise similarity between the nominal response functions of channel i and channel j. The larger the value, the stronger the correlation between the nominal response functions of the two channels and the more obvious the spectral overlap; The smaller the value, the higher the degree of separation between the two channels.

7. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S3, the calculation method for the overall separability evaluation quantity of the three channels is as follows: A weighted Gram matrix G is constructed based on the weighted correlation coefficients between each channel. The elements of matrix G are defined as G... ij : ； Where, when i=j, G ij =1 represents the diagonal element, indicating the normalized correlation between each channel and itself; Off-diagonal elements represent the weighted correlation between different channels; The condition number of the eigenvalues ​​of the weighted Gram matrix G for: ； in, and Let G and G represent the largest and smallest eigenvalues ​​of the weighted Gram matrix G, respectively. The condition number It is a three-channel overall separability evaluation quantity, used to characterize the overall joint separability of the three-channel nominal response functions and the numerical stability of the subsequent solution process; The larger the value, the stronger the overall correlation and the worse the separability among the three-channel nominal response functions, and the more sensitive the subsequent optimization solution is to noise and parameter perturbations. The smaller the value, the better the overall separability of the three-channel nominal response function, and the higher the stability of subsequent solutions.

8. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S4, the process of constructing the comprehensive optimization objective function is as follows: Step 4.1: Construct channel energy balance evaluation indicators : ； The channel energy balance evaluation index Used to characterize the degree of deviation between the channel energy allocation corresponding to the current three-channel nominal response function and the preset target energy ratio; The smaller the value, the closer the energy distribution of the three channels is to the preset target ratio, and the lower the degree of imbalance in the response of the three channels during subsequent color restoration; Step 4.2: Construct an overall three-channel separability evaluation index : ； The larger the value, the stronger the overall correlation and the worse the joint separability among the three-channel nominal response functions; The smaller the value, the better the overall separability of the three-channel nominal response function, and the higher the stability of subsequent solutions; Step 4.3: Construct the curve morphology of the transmittance function and the evaluation index of process feasibility. : ； in, and They are respectively and Sub-item weight coefficients For edge shape penalty term, This is a penalty item for bypassing the zone; The edge shape penalty item This term is used to characterize the deviation between the transition characteristics of the transmittance function of each channel filter at the passband edge and the edge morphology achievable by a preset process; the out-of-band transmittance penalty term. Used to characterize the residual transmittance of each channel filter outside the nominal passband; Step 4.4, construct the comprehensive optimization objective function: Step 4.4.1, Define the parameter vector to be optimized. : ； Step 4.4.2, in the parameter vector to be optimized Above, construct a comprehensive optimization objective function. : ； in, , , and These are the total relative fluxes of the three channels. Channel energy balance evaluation indicators Three-channel overall separability evaluation index The curve morphology of the transmittance function and the evaluation index of process feasibility Weighting coefficients; The weighting coefficient , , and The parameters are set according to the total throughput priority, channel separation requirements, energy balance requirements and process implementation requirements, and are adjusted based on experimental experience, numerical simulation or verification results; The comprehensive optimization objective function Constructed in a minimal form, where, The form of a negative sign is used to enter the objective function, which is used to drive the maximization of the total relative flux of the three channels within the minimization framework; , as well as The penalty term is incorporated into the comprehensive optimization objective function. .

9. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S4, the solution process for the comprehensive optimization objective function is as follows: S4.1, Perform a global search under constraints to obtain a candidate parameter vector as initial values: Within the manufacturable design domain, a global search is performed using a genetic algorithm or a differential evolution algorithm to obtain a set of candidate parameters that satisfy the constraints and the corresponding candidate comprehensive optimization objective function values. The global search is used to search for candidate parameter vectors that satisfy the constraints in the parameter space, reducing the sensitivity of the solution process to the selection of initial values, and providing initial parameters for subsequent local fine-tuning. S4.2, perform local fine-tuning of the comprehensive optimization objective function: While maintaining the manufacturability constraints, material boundary constraints, system boundary constraints, and channel spacing constraints given in S2, the comprehensive optimization objective function is locally fine-tuned to obtain the optimal parameter vector. Optimized three-channel filter transmittance function and the optimized three-channel nominal response function ; The local fine-tuning employs a local optimization method based on sequential quadratic programming or trust region-based constraint optimization to further improve the accuracy of parameter solving and enable the comprehensive optimization objective function to converge to the optimal solution within the current constraint domain.

10. The RGB filter design method for low-light night vision according to claim 1, characterized in that, In S5, the final qualitative process of the transmittance function of the three-channel filter is as follows: Step 5.1, perform peak-limited calibration on the R and G channels: Define the transmittance function of the calibrated R-channel filter respectively. and the transmittance function of the G-channel filter : ； ； in, and They are respectively and The scaling factor satisfies: ； ； This guarantees: ； ； Among them, the upper limit of peak transmittance is ; Step 5.2, Determining the energy state of channel B: The optimized three-channel nominal response function based on the S4 output Define the optimized channel energy of the i-th channel as: : ； Right now: ； For channel B, if the channel energy E is optimized... B After optimization of the G channel, the channel energy E G satisfy: ； Then it is determined that channel B has insufficient energy; among which, The preset energy ratio threshold is used to characterize the minimum allowable energy ratio of the B channel relative to the G channel; Step 5.3: Perform limiting compensation processing on channel B to obtain the qualitative transmittance function of channel B filter; When it is determined that the B channel has insufficient energy, and the B channel compensation does not cause the overall separability evaluation value of the three channels to exceed the preset threshold, the transmittance function of the optimized B channel filter is... After performing limited compensation processing, the transmittance function of the B-channel filter after finalization is obtained. ; The restricted compensation process improves the effective spectral energy of the B channel by restrictively adjusting the local transmittance distribution and edge transition shape within the effective passband of the B channel, while controlling the increase in the additional overlap between the B channel and the G channel. The restricted compensation process simultaneously satisfies the following conditions: First, it does not disrupt the preset spectral boundaries of channel B at the shortwave and longwave ends; Second, do not exceed the peak transmittance limit. and minimum slope constraint at the edge; Third, prevent the out-of-band transmittance integral value from exceeding the preset upper limit; Fourth, the overall separability evaluation value of the three channels should not exceed the preset threshold; If the B-channel energy deficiency judgment is not triggered, the transmittance function of the finalized B-channel filter will be... : ； If the insufficient energy determination of channel B is triggered, then under the premise of satisfying the constraints, ... After local compensation adjustments, the transmittance function of the B-channel filter after finalization is obtained. ; Step 5.4, output the final RGB filter transmittance data: Final R-channel filter transmittance function and the transmittance function of the G-channel filter Obtained from the peak-limited calibration in step 5.1; Final B-channel filter transmittance function It is obtained from the B-channel limitation compensation process in step 5.3.

Images