TDI CCD wake imaging parameter optimization method and system based on stochastic optimization

Abstract

The invention provides a TDI CCD wake imaging parameter optimization method and system based on stochastic optimization, and belongs to the technical field of remote sensing imaging. The method comprises the following steps: carrying out imaging modeling on a TDI CCD imaging system to obtain two-stage imaging models; in the first stage, the process of outputting electrons by pixels is modeled into a random variable model with probability distribution being Poisson distribution; in the second stage, modeling a process of reading a pixel digital signal value into a Gaussian random process model; building a stochastic optimization model based on the two-stage imaging model, and modeling an optimization problem as a Gaussian process stochastic optimization problem based on input uncertainty; converting a Gaussian process stochastic optimization problem based on input uncertainty into a sample mean value approximation problem by using a stochastic simulation method; and solving the problem of sample mean value approximation by adopting a simulated annealing method to obtain parameter configuration when the DN value is maximum. The problem that a traditional imaging system is not clear in dark wake imaging is solved simply and effectively.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing imaging, and in particular to a TDI CCD trail imaging parameter optimization method and system based on random optimization. Background Art

[0002] Ship wakes are important detection targets in marine remote sensing. By analyzing ship wakes, we can obtain characteristic information such as the type, location, speed, and heading of the ship target. This information is of great significance for maritime traffic safety, military reconnaissance, and marine environmental protection. Conventionally, wake signals are relatively weak, and the imaging signal-to-noise ratio is low. As a result, existing remote sensing imaging systems often produce images of wake targets with low contrast and blurred details. Therefore, obtaining clear images of wake targets is of great significance.

[0003] Typical methods for acquiring clear wake images are mainly divided into image processing-based methods and imaging system optimization-based methods. Image processing-based methods mainly focus on enhancing wakes through image denoising and enhancement methods. However, if the wake signal is too weak and affected by noise, the image signal obtained by the imaging system often cannot display the wake information. Imaging system optimization-based methods mainly use prior knowledge to analyze the signal-to-noise ratio of the image, thereby obtaining the relationship between indicators such as the signal-to-noise ratio and the imaging system parameters, and then provide the imaging system optimization design results. However, such methods ignore the random influence of the signal and noise, and have certain performance limitations.

[0004] Therefore, a parameter optimization method that can achieve random optimization is needed to improve the clarity of wake imaging capabilities. Summary of the Invention

[0005] In view of this, the present invention provides a TDI CCD trail imaging parameter optimization method and system based on random optimization. By analyzing the characteristics of the TDICCD (Time Delay and Integration Charge Coupled Devices) imaging system, the full-link imaging effect of the system is clarified, and how to select imaging parameters to achieve the best imaging performance for faint trails is studied, so as to solve the problem of unclear trail imaging through existing remote sensing imaging systems.

[0006] To this end, the present invention provides the following technical solutions:

[0007] The present invention discloses a TDI CCD trail imaging parameter optimization method based on random optimization, comprising the following steps:

[0008] The imaging model of the TDI CCD imaging system is constructed to obtain a two-stage imaging model. In the first stage, the process of pixel outputting electrons is modeled as a random variable model with a Poisson distribution probability distribution. In the second stage, the process of reading the pixel digital signal value is modeled as a Gaussian random process model.

[0009] A stochastic optimization model is constructed based on the two-stage imaging model. The stochastic optimization model takes maximizing the readout pixel digital signal value DN of the imaging system as the optimization goal, uses the optical system F number F, the integral series M, and the quantization bit n as optimization parameters, and models the optimization problem as a Gaussian process stochastic optimization problem based on uncertain input, wherein the output of the random variable model established in the first stage serves as the input of the Gaussian process model established in the second stage;

[0010] The random simulation method is used to transform the Gaussian process random optimization problem based on uncertain input into the problem of sample mean approximation.

[0011] The simulated annealing method is used to solve the sample mean approximation problem, and the optical system F number F, the integration series M and the quantization bit number n when the DN value is maximized are obtained.

[0012] Furthermore, the first stage models the process of pixel outputting electrons as a random variable model with a probability distribution of Poisson distribution, including:

[0013] In the process of pixel outputting electrons, the number of output electrons EN is expressed as:

[0014]

[0015] Among them, EN is the number of pixel output electrons; N ei Indicates the number of photogenerated electrons of the sensor at level i integration, N shoti N represents the equivalent electron number of the sensor's grain noise at level i integration. shoti The probability distribution satisfies (N shoti +N ei )~Pois(N ei );N darki N represents the dark current noise equivalent electron number of the i-th level sensor, darki The probability distribution is N darki ~Pois(σ dark 2 );

[0016] The process of pixel outputting electrons is modeled as a random variable model with a probability distribution of Poisson distribution, which can be expressed as:

[0017]

[0018] Among them, τ a is the atmospheric transmittance, τ o is the total transmittance of the optical system, R is the average spectral responsivity, T int is the integration time, M max is the maximum integral series, C is the charge conversion factor, ξ transfer is the total charge transfer efficiency, N fullwell is the full well charge of the sensor, σ dark is the dark current noise equivalent electron number, σ rms is the readout noise equivalent electron number, R r is the target wake reflection resolution, and B is the ambient irradiance.

[0019] Furthermore, the second stage models the process of reading pixel digital signal values ​​as a Gaussian random process model, including:

[0020] In the process of reading the pixel digital signal value, the output DN value is expressed as:

[0021]

[0022] Among them, N rsm N is the equivalent number of electrons representing the readout noise. srm The probability distribution is N srms ~N(0,σ rms 2 );

[0023] The process of reading pixel digital signal values ​​is modeled as a Gaussian random process, which can be expressed as:

[0024] y GP (x) = Ψ(x) + ε(x);

[0025] Where Ψ(x) is expressed as ε(x) represents the readout noise, which obeys Normal distribution.

[0026] Furthermore, the stochastic optimization model is:

[0027]

[0028]

[0029] E[y GP (x)]<2 n ;

[0030] Where u represents the input vector, u = [F, M, n], and U represents the value range of u.

[0031] Furthermore, the method of using random simulation transforms the Gaussian process random optimization problem based on uncertain input into a sample mean approximation problem, including:

[0032]

[0033]

[0034]

[0035] where y GPj (x) represents the DN value output in the jth experiment, and N represents the number of experiments to approximate the sample mean.

[0036] Furthermore, the simulated annealing method is used to obtain the optical system F number F, the integral series M and the quantization bit number n when the DN value is maximized, including:

[0037] S1 sets the initial temperature t0 and the stop temperature t f , current temperature t current Equal to t0; decay rate r, upper limit of iteration number n markov , output judgment parameter D;

[0038] S2 sets the initial value u0 of the decision variable and generates the current initial solution u current =u0, as the current optimal solution u best , start iterative operation;

[0039] S3 loops through the iterative calculation process until the upper limit of the number of iterations n is reached markov ;

[0040] At this time, determine whether the output conditions are met, and if so, output the result;

[0041] If it is not satisfied, lower the temperature and set t current Equal to r*t current , continue to start a new round of iteration until the output condition is met or the current temperature t current Less than the stop temperature t f , output the result; the result includes the optical system F, the number of integral series, and the number of quantization bits when the DN value is maximized.

[0042] Furthermore, the iterative calculation process is:

[0043] For the current solution u current Add a small perturbation to generate a new solution u new ;

[0044] For g(x,u new ) to judge, greater than g(x,u best ), then accept the new solution, that is, let ucurrent Equal to u new ,u best Equal to u new ;

[0045] g(x,u new ) is less than or equal to g(x,u best ), then the new solution is accepted according to the Metropolis criterion; that is, when the new solution is not better than the optimal solution, the probability is Accept new interpretations.

[0046] Furthermore, the output condition is that the difference between the last obtained local optimal value and the current optimal value is less than D.

[0047] The present invention also discloses a TDI CCD trail imaging parameter optimization system based on random optimization, comprising: a data input module, an imaging modeling module, an optimization modeling module, a random simulation module and a solution module;

[0048] The data input module is used to input the basic parameters and parameters to be optimized required by the imaging modeling module;

[0049] The imaging modeling module is used to perform imaging modeling on the TDI CCD imaging system according to the parameters transmitted by the data input module, thereby obtaining a two-stage imaging model. In the first stage, the process of pixel outputting electrons is modeled as a random variable model with a probability distribution of Poisson distribution; in the second stage, the process of reading pixel digital signal values ​​is modeled as a Gaussian random process model.

[0050] The optimization modeling module is used to construct a stochastic optimization model based on the two-stage imaging model obtained by the imaging modeling module. The stochastic optimization model takes maximizing the readout pixel digital signal value DN of the imaging system as the optimization goal, uses the optical system F number F, the integral series M, and the quantization bit n as optimization parameters, and models the optimization problem as a Gaussian process stochastic optimization problem based on uncertain input, wherein the output of the random variable model established in the first stage serves as the input of the Gaussian process model established in the second stage;

[0051] The stochastic simulation module uses a stochastic simulation method to convert the Gaussian process stochastic optimization problem based on input uncertainty modeled by the optimization modeling module into a sample mean approximation problem;

[0052] The solving module adopts simulated annealing method to solve the problem after the random simulation module conversion, and obtains the optical system F number F, the integral series M and the quantization bit number n when the DN value is maximized.

[0053] Advantages and positive effects of the present invention:

[0054] The present invention provides a TDICCD wake imaging parameter optimization method based on random optimization, which solves the problem of unclear imaging of faint wakes in traditional imaging systems. The system's faint wake imaging capability is optimized based on the modeling of the TDICCD imaging process and the random optimization method. The method is simple, has a small amount of calculation, improves the system's ability to clearly image wakes, and can provide guidance for the parameter design of imaging systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0056] Figure 1 Flowchart of a TDICCD wake imaging parameter optimization method based on random optimization in an embodiment of the present invention;

[0057] Figure 2 4 is a structural diagram of a TDICCD trail imaging parameter optimization system based on random optimization in an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0059] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0060] The present invention provides a TDICCD trail imaging parameter optimization method based on random optimization. By simulating TDICCD full-link imaging, the system characteristics and the statistical characteristics of signals and noise in the imaging link are analyzed, and the distribution characteristics of digital signals are obtained. Then, a random optimization model is established based on the digital signal output requirements. By solving the model, the optimal imaging parameters are obtained, providing theoretical and practical guidance for the parameter design of the imaging system.

[0061] like Figure 1 As shown in FIG, a TDICCD wake imaging parameter optimization method based on random optimization in an embodiment of the present invention specifically includes the following steps:

[0062] Step a: input the basic parameters and parameters to be optimized required by the TDI CCD imaging system;

[0063] In this embodiment, the basic parameters set include: atmospheric transmittance τ a is 1, and the total transmittance of the optical system is τ o is 0.9, and the average spectral responsivity is R10 5 , the integration time is T int 3*10 -4 , the maximum integral series is M max is 96, and the charge conversion factor C is 10 -5 , the total charge transfer efficiency is ξ transfer is 1, the sensor full well charge number is N fullwell is 10 5 , the dark current noise equivalent electron number is σ dark is 100, and the readout noise equivalent electron number is σ rms is 100, and the target trail reflection resolution is R r is 0.0001, and the ambient irradiance is B is 1;

[0064] The parameters to be optimized include: the optical system F number is F, which is 9, the integration level is M, which is 8, and the number of quantization bits is n, which is 8.

[0065] Step b, modeling the TDI CCD imaging system;

[0066] The imaging process of TDI CCD is decomposed into two stages, and the imaging system is modeled as a two-stage imaging model. The first stage is the process of pixel outputting electrons, and the second stage is the process of reading the pixel digital signal value DN.

[0067] (1) Model the process of pixel outputting electrons, specifically:

[0068] In the process of pixel outputting electrons, the number of output electrons EN can be expressed as:

[0069]

[0070] Among them, EN is the number of pixel output electrons; N ei Indicates the number of photogenerated electrons of the sensor at level i integration, N shoti N represents the equivalent electron number of the sensor's grain noise at level i integration. shoti The probability distribution satisfies (N shoti +N ei )~Pois(N ei );N darki N represents the dark current noise equivalent electron number of the i-th level sensor, darki The probability distribution is N darki ~Pois(σ dark 2 ).

[0071] The process of pixel outputting electrons is modeled as a random variable model with a Poisson distribution probability distribution, which can be expressed as:

[0072]

[0073] (2) Modeling the process of reading pixel digital signal values;

[0074] In the process of reading the pixel digital signal value, the output DN value can be expressed as:

[0075]

[0076] Among them, N rsm N is the equivalent number of electrons representing the readout noise. srm The probability distribution is N srms ~N(0,σ rms 2 ).

[0077] The above process of reading pixel digital signal values ​​is modeled as a Gaussian random process, which can be expressed as:

[0078] y GP (x)=Ψ(x)+ε(x)

[0079] Among them, Ψ(x) can be expressed as

[0080] ε(x) represents the readout noise, which obeys Normal distribution.

[0081] Step c: Establish a random optimization model.

[0082] Based on the two-stage imaging model described above, maximizing the imaging system's DN output is the optimization objective, with F, M, and n as optimization parameters. The optimization problem is modeled as a Gaussian process stochastic optimization problem with uncertain inputs. The output of the random variable model established in the first stage serves as the input to the Gaussian process model established in the second stage. The specific process is as follows:

[0083]

[0084]

[0085] E[y GP (x)<2 n ;

[0086] Where u represents the input vector, u = [F, M, n], and U represents the value range of u. In this embodiment, the value of u is: [[5, 6, 7, 8, 9, 10], [8, 16, 32, 48, 64, 96], [8, 10, 12]].

[0087] Step d: Use the random simulation method to transform the original problem into a sample mean approximation problem.

[0088] Using the sample mean approximation method, the original problem input uncertain Gaussian process optimization problem is converted into a deterministic optimization problem, which can be expressed as:

[0089]

[0090]

[0091]

[0092] where y GPj (x) represents the DN value output in the jth experiment, and N represents the number of experiments for approximating the sample mean, which is 1000 in this embodiment.

[0093] Step e: using simulated annealing to obtain the parameter configuration of the DN maximization problem.

[0094] Specifically, the simulated annealing method is used to solve the model and obtain the optical system F number, integration level, and quantization bit number when the DN value is maximized. The steps are as follows:

[0095] Step 1: Set the initial temperature t0 to 1000 and the stop temperature t f for 3 , current temperature t current Equal to t0, the decay rate r is 0.9, and the upper limit of the number of iterations is n markov is 100, the output judgment parameter D is 3;

[0096] Step ②, current temperature t current Equal to t0.

[0097] Step 3: Set the initial value of the decision variable u0 = [9, 8, 8] and generate the current initial solution u current =u0, and is considered to be the current optimal solution u best , start iterative operation;

[0098] Step ④: For the current solution u current Add a small perturbation to generate a new solution u nerw ;

[0099] Step 5: If g(x,u new ) is greater than g(x,u best ), then accept the new solution, that is, let u current Equal to u new ,u best Equal to u new ; If g(x,u new ) is not greater than g(x,u best ), then the new solution is accepted according to the Metropolis criterion, where the Metropolis criterion is that when the new solution is not better than the optimal solution, the probability is accept new interpretations;

[0100] Step 6: Determine whether the number of iterations reaches the upper limit n markov If it is not reached, continue to repeat steps ③ and ④;

[0101] Step 7: If the number of iterations reaches the upper limit, determine whether the output condition is met. The output condition is that the difference between the last local optimal value and the current optimal value is less than 3. If the output condition is not met, reduce the temperature and set t current Equal to r*t current , continue to start a new round of iteration;

[0102] Step ⑧: If the output condition is met or the temperature drops to the current temperature t current Less than the stop temperature t f , then output the result. Finally, the optical system F number that maximizes the DN value is 5, the integration level is 96, and the quantization bit is 12.

[0103] Corresponding to the TDICCD wake imaging parameter optimization method based on random optimization in the above embodiment, the present invention also provides a TDICCD wake imaging parameter optimization system based on random optimization, which is used to execute the TDICCD wake imaging parameter method described in the above method embodiment.

[0104] like Figure 2As shown in the system structure diagram of the present invention, the TDI CCD trail imaging parameter optimization system based on random optimization includes the following modules: data input module, imaging modeling module, optimization modeling module, random simulation module, and solution module.

[0105] Data input module: The data input module is configured to input the basic parameters and parameters to be optimized required by the imaging modeling module. The basic parameters to be set include: atmospheric transmittance τ a is 1, and the total transmittance of the optical system is τ o is 0.9, and the average spectral responsivity is R10 5 , the integration time is T int 3*10 -4 , the maximum integral series is M max is 96, and the charge conversion factor C is 10 -5 , the total charge transfer efficiency is ξ transfer is 1, the sensor full well charge number is N fullwell is 10 5 , the dark current noise equivalent electron number is σ dark is 100, and the readout noise equivalent electron number is σ rms is 100, and the target trail reflection resolution is R r is 0.0001, and the ambient irradiance is B is 1; the parameters to be optimized include: the optical system F number is F is 9, the integration level is M is 8, and the number of quantization bits is n is 8.

[0106] Imaging Modeling Module: This module is configured to establish a TDI CCD imaging model based on the parameters transmitted by the data input module. This established model is a two-stage imaging model. The first stage is the process of pixel electronic output, and the second stage is the process of reading the pixel digital signal value DN. The imaging modeling module transmits this established two-stage imaging model to the optimization modeling module.

[0107] Optimization Modeling Module: The optimization modeling module is configured to transform the model transmitted by the imaging modeling module into a stochastic optimization problem solvable using an optimization algorithm, with maximizing the DN output of the imaging system as the optimization objective. The optimization parameters are u = [F, M, n], and the range of the optimization parameters is U = [[5, 6, 7, 8, 9, 10], [8, 16, 32, 48, 64, 96], [8, 10, 12]]. The optimization problem is modeled as a stochastic optimization problem based on a Gaussian process with uncertain inputs, wherein the output of the random variable model established in the first stage serves as the input of the Gaussian process model established in the second stage. The optimization modeling module transmits the above-established stochastic optimization problem based on a Gaussian process with uncertain inputs to the stochastic simulation module.

[0108] Stochastic Simulation Module: This module is configured to use stochastic simulation to convert the stochastic optimization problem transmitted from the optimization modeling module into a sample mean approximation problem. This method converts the original uncertain Gaussian process optimization problem into a deterministic optimization problem. The number of sample mean approximation experiments is N, which is 1000 in this example. The stochastic simulation module transmits this converted sample mean approximation problem to the solution module.

[0109] Solving module: The solving module is constructed to use simulated annealing algorithm to solve the sample mean approximation problem transmitted by the random simulation module. In the process of using simulated annealing method, basic parameters need to be set, including setting the initial temperature t0 to 1000, the stopping temperature t f is 3, the decay rate r is 0.9, and the upper limit of the number of iterations n markov =100, the output judgment parameter D is 3 and the initial value of the decision variable u0=[9,8,8]. In this embodiment, the solution module outputs the optical system F number when the DN value is maximized as 5, the integration level is 96 and the quantization bit number is 12.

[0110] Compared with the prior art, the present invention has the following advantages:

[0111] The present invention provides a TDICCD wake imaging parameter optimization method based on random optimization, which solves the problem of imaging systems detecting weak wakes. Based on the modeling of the TDICCD imaging process and the random optimization method, the imaging capability of the imaging system for weak wakes is optimized. The method is simple, has a small amount of calculation, ensures the presence of wake signals in the image, improves the system's imaging capability for wakes, and can be used to provide guidance for the parameter design of imaging systems.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A TDICCD trail imaging parameter optimization method based on random optimization, characterized in that: The following steps are involved: The imaging model of TDICCD imaging system is carried out to obtain a two-stage imaging model; in the first stage, the process of pixel outputting electrons is modeled as a random variable model with a probability distribution of Poisson distribution; in the second stage, the process of reading out the pixel digital signal value is modeled as a Gaussian random process model; A random optimization model is constructed based on the two-stage imaging model. The random optimization model takes maximizing the readout pixel digital signal value DN of the imaging system as the optimization target, takes the optical system F number F, the integral series M and the quantization bit number n as the optimization parameters, and models the optimization problem as a Gaussian process random optimization problem based on uncertain input, wherein the output of the random variable model established in the first stage is used as the input of the Gaussian process model established in the second stage; The random simulation method is used to transform the Gaussian process random optimization problem based on uncertain input into the problem of sample mean approximation. The simulated annealing method is used to solve the sample mean approximation problem, and the optical system F number F, the integral series M and the number of quantization bits n when the DN value is maximized are obtained.

2. The TDICCD trail imaging parameter optimization method and system based on random optimization according to claim 1, characterized in that: The first stage models the process of pixel outputting electrons as a random variable model with a probability distribution of Poisson distribution, including: In the process of pixel outputting electrons, the number of output electrons EN is expressed as: Among them, EN is the number of pixel output electrons; N ei It represents the number of photogenerated electrons of the sensor at level i integration. N shoti N represents the equivalent electron number of the sensor's grain noise at level i integration. shoti The probability distribution satisfies (N shoti +N ei )~Pois(N ei );N darki N represents the dark current noise equivalent electron number of the i-th sensor. darki The probability distribution is N darki ~Pois(σ dark 2 ); The process of pixel outputting electrons is modeled as a random variable model with a probability distribution of Poisson distribution, which can be expressed as: Among them, x is the output electron, τ a is the atmospheric transmittance, τ o is the total transmittance of the optical system, R is the average spectral responsivity, T int is the integration time, M max is the maximum integral series, C is the charge conversion factor, ξ transfer is the total charge transfer efficiency, N fullwell is the sensor full well charge number, σ dark is the dark current noise equivalent electron number, σ rms is the readout noise equivalent electron number, R r is the target wake reflection resolution, and B is the ambient irradiance.

3. The TDICCD trail imaging parameter optimization method and system based on random optimization according to claim 2, characterized in that: The second stage models the process of reading out the pixel digital signal value as a Gaussian random process model, including: In the process of reading the pixel digital signal value, the output DN value is expressed as: Among them, N rsm N is the equivalent number of electrons of the readout noise. srm The probability distribution is N srms ~N(0,σ rms 2 ); The process of reading out the pixel digital signal value is modeled as a Gaussian random process, expressed as: y GP (x)=Ψ(x)+ε(x); Where Ψ(x) is expressed as ε(x) represents the readout noise, which is The normal distribution of .

4. The TDICCD trail imaging parameter optimization method and system based on random optimization according to claim 3, characterized in that: The stochastic optimization model is: Hey GP (x)]<2 n ; Wherein u represents the input vector, u=[F, M, n], and U represents the value range of u.

5. The TDICCD trail imaging parameter optimization method and system based on random optimization according to claim 1, characterized in that: The method of using random simulation transforms the Gaussian process random optimization problem based on uncertain input into a sample mean approximation problem, including: where y GPj (x) represents the DN value output in the jth experiment, N represents the number of experiments for approximating the sample mean, and g(x, u) is the objective function.

6. The method and system for optimizing TDICCD trail imaging parameters based on random optimization according to claim 5, characterized in that: The method of using the simulated annealing method to obtain the optical system F number F, the integral series M and the quantization bit number n when the DN value is maximized includes: S1 sets the initial temperature t0 and the stop temperature t f , current temperature t current Equal to t0; decay rate r, upper limit of iteration number n markov , output judgment parameter D; S2 sets the initial value u0 of the decision variable and generates the current initial solution u current =u0, as the current optimal solution u best , start iterative operation; S3 loops through the iterative calculation process until the upper limit of the number of iterations n is reached markov ; At this time, determine whether the output conditions are met, and if so, output the result; If it is not satisfied, lower the temperature and set t current Equal to r*t current , and continue to start a new round of iteration until the output condition is met or the current temperature t current Less than the stop temperature t f , output the result; the result includes the optical system F, the number of integral series, and the number of quantization bits when the DN value is maximized.

7. The method and system for optimizing TDICCD trail imaging parameters based on random optimization according to claim 6, characterized in that: The iterative operation process is: For the current solution current Add a small disturbance to generate a new solution u new ; For g(x,u new ) is judged to be greater than g(x,u best ), then accept the new solution, that is, let u current Equal to u new ,u best Equal to u new ; g(x,u new ) is less than or equal to g(x,u best ), then accept the new solution according to the Metropolis criterion; That is, when the new solution is not better than the optimal solution, the probability is Accept new interpretations.

8. The method and system for optimizing TDICCD trail imaging parameters based on random optimization according to claim 6, characterized in that: The output condition is that the difference between the last obtained local optimal value and the current optimal value is less than D.

9. A TDICCD trail imaging parameter optimization system based on random optimization, characterized in that: include: Data input module, imaging modeling module, optimization modeling module, random simulation module and solution module; The data input module is used to input the basic parameters and parameters to be optimized required by the imaging modeling module; The imaging modeling module is used to perform imaging modeling on the TDI CCD imaging system according to the parameters transmitted by the data input module to obtain a two-stage imaging model; in the first stage, the process of pixel outputting electrons is modeled as a random variable model with a probability distribution of Poisson distribution; in the second stage, the process of reading out the pixel digital signal value is modeled as a Gaussian random process model; The optimization modeling module is used to construct a random optimization model based on the two-stage imaging model obtained by the imaging modeling module. The random optimization model takes maximizing the readout pixel digital signal value DN of the imaging system as the optimization target, takes the optical system F number F, the integral series M and the quantization bit number n as the optimization parameters, and models the optimization problem as a Gaussian process random optimization problem based on uncertain input, wherein the output of the random variable model established in the first stage is used as the input of the Gaussian process model established in the second stage; The random simulation module uses a random simulation method to convert the Gaussian process random optimization problem based on input uncertainty modeled by the optimization modeling module into a sample mean approximation problem; The solving module uses simulated annealing method to solve the problem after the random simulation module conversion, and obtains the optical system F number F, the integral series M and the number of quantization bits n when the DN value is maximized.

Images