Methods, apparatus, equipment, media, and program products for reconstructing damaged satellite image data based on simulated annealing genetic algorithm.

Abstract

This disclosure provides a method, apparatus, device, medium, and program product for reconstructing damaged satellite image data based on simulated annealing genetic algorithm. Applied to the field of satellite image data processing technology, the method includes: encoding the original image based on initialization parameters to obtain a first population consisting of multiple first individuals; calculating the fitness of each first individual in the first population; randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual; calculating the fitness of the second individual in the second population, and selecting the optimal individual in the second population based on the fitness; adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population; performing a cooling operation during each iteration, and in response to the current temperature meeting a preset condition, using the optimal individual in the second population as the reconstructed image of the image to be processed.

Description

Technical Field

[0001] This disclosure relates to the field of satellite image data processing technology, and more specifically to a method, apparatus, equipment, medium, and program product for reconstructing damaged satellite image data based on simulated annealing genetic algorithm. Background Technology

[0002] In space exploration missions, especially in deep space exploration and satellite remote sensing imaging, the integrity of image data is directly related to the success of scientific research and missions. However, due to various factors, such as noise interference, signal attenuation, or space radiation during data transmission, image data is often damaged or lost. Especially in the transmission of high-precision remote sensing images, the integrity of image data is often difficult to guarantee due to the large capacity of satellite images, long distances, and limitations of communication equipment.

[0003] Image data reconstruction technology aims to restore the original information of an image as much as possible by recovering damaged parts, which is of great significance for aerospace image analysis and post-processing. Existing image reconstruction methods mainly include interpolation methods, model-based methods, and deep learning-based reconstruction methods.

[0004] However, most existing image reconstruction techniques employ a single method for image restoration. For example, interpolation relies on simple mathematical models, which, while easy to implement, yield poor reconstruction results in scenarios with complex image damage or severe loss of structural information, failing to meet the requirements for high-quality satellite image restoration. Model-based reconstruction methods depend on prior knowledge of the image and typically perform well with minor image damage, but their accuracy is significantly insufficient when the damaged area is large or the image structure is complex, making them unsuitable for reconstructing complex satellite images. While deep learning-based image reconstruction methods offer advantages in restoration quality, their training process relies on large amounts of high-quality data and incurs significant computational costs, making them difficult to apply effectively in scenarios with high real-time and resource constraints, such as aerospace missions. Summary of the Invention

[0005] This disclosure provides a method, apparatus, device, and storage medium for reconstructing damaged satellite image data based on simulated annealing genetic algorithm.

[0006] According to the first aspect of this disclosure, a method for reconstructing damaged satellite image data based on simulated annealing genetic algorithm is provided, comprising: encoding the original image based on initialization parameters to obtain a first population consisting of multiple first individuals, each individual being a pre-reconstructed image corresponding to the original image, wherein the encoding method of the damaged area in the original image is different in different individuals; performing N iterations with an initial temperature as the starting temperature, where N is a positive integer; the iteration process includes: calculating the fitness of each first individual in the first population, the fitness being used to represent the closeness between the pre-reconstructed image and the original image; randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual; calculating the fitness of the second individual in the second population, and selecting the optimal individual in the second population based on the fitness; adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population; performing a cooling operation in each iteration, and determining the optimal individual in the second population as the reconstructed image of the image to be processed in response to the current temperature meeting a preset condition.

[0007] According to embodiments of this disclosure, calculating the fitness of each first individual in a first population includes performing the following operations on each first individual: determining the mean square error, structural similarity index, and image quality index of the original image and the pre-reconstructed image; determining the fitness of the first individual based on the mean square error, structural similarity index, and image quality index; and stretching the fitness of the first individual based on the temperature of the first population and the fitness of multiple first individuals in the first population to adjust the relative position of the first individual in the first population.

[0008] According to embodiments of this disclosure, two first individuals are randomly selected from an initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individuals. This includes: classifying the first individuals based on preset conditions and the stretched fitness of each first individual to obtain Class I individuals and Class II individuals; for Class I individuals, directly copying the first individual to the second population; for Class II individuals, randomly selecting two first individuals from the Class II individuals for adaptive crossover and adaptive mutation to obtain sub-individuals of the Class II individuals; adding at least some of the sub-individuals of the Class II individuals to the second population as second individuals; wherein, adaptive crossover is used to exchange pixel blocks in a certain pixel region between two first individuals; and adaptive mutation is used to randomly change a certain pixel block in the sub-individuals.

[0009] According to embodiments of this disclosure, the method further includes: calculating the crossover probability and mutation probability corresponding to each first individual based on the fitness statistics in the first population and the fitness of each first individual; using the crossover probability to adjust the frequency of the crossover operation and the mutation probability to adjust the frequency of the mutation operation; and performing the crossover operation or mutation operation on the individual based on the crossover probability and mutation probability of the first individual.

[0010] According to embodiments of this disclosure, adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population includes: randomly modifying the parameters of the optimal individual to obtain a new individual corresponding to the optimal individual; calculating the fitness of the new individual and comparing the fitness of the new individual with the fitness of the optimal individual; replacing the optimal individual in the second population with the new individual if the fitness of the new individual is not less than the fitness of the optimal individual; and accepting the new individual with a preset probability if the fitness of the new individual is less than the fitness of the optimal individual.

[0011] According to embodiments of this disclosure, the method further includes: gradually performing a cooling operation during the iteration process; when the current temperature is lower than the target temperature, using the best individual in the second population as the reconstructed image of the image to be processed; and continuing the iteration process when the current temperature is not lower than the target temperature.

[0012] According to a second aspect of this disclosure, a satellite image damaged data reconstruction device based on simulated annealing genetic algorithm is provided, comprising: an encoding module, used to encode the original image based on initialization parameters to obtain a first population consisting of multiple first individuals, each individual being a pre-reconstructed image corresponding to the original image, wherein the encoding method of the damaged area in the original image is different in different individuals; an iteration module, used to perform N iterations with an initial temperature as the starting temperature, where N is a positive integer; the iteration process includes: calculating the fitness of each first individual in the first population, the fitness being used to represent the closeness between the pre-reconstructed image and the original image; randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual; calculating the fitness of the second individual in the second population, and selecting the optimal individual in the second population based on the fitness; adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population; and a determination module, used to perform a cooling operation in each iteration, and in response to the current temperature meeting a preset condition, determining the optimal individual in the second population as the reconstructed image of the image to be processed.

[0013] According to a third aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the methods described above.

[0014] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided that stores a computer program or instructions thereon, which, when executed by a processor, implement the steps of the method described above.

[0015] According to a fifth aspect of this disclosure, a computer program product is provided, comprising a computer program or instructions that, when executed by a processor, implement the steps of the method described above.

[0016] The satellite image damaged data reconstruction method based on simulated annealing genetic algorithm disclosed herein has at least the following beneficial effects:

[0017] By employing adaptive crossover and adaptive mutation, combined with temperature control from simulated annealing, the algorithm dynamically switches between global search and local fine-tuning, avoiding getting trapped in local optima. Perturbations are added to the optimal individuals to help them escape local optima and explore a better solution space, achieving synergy between global optimization and local search, thus avoiding local optima. Elite preservation and dynamic adjustment strategies effectively improve convergence performance and population diversity. Fitness stretching and temperature control balance image reconstruction accuracy and efficiency, maintaining stable convergence while possessing a strong ability to escape local optima. Attached Figure Description

[0018] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0019] Figure 1 A flowchart illustrating a method for reconstructing damaged satellite image data according to an embodiment of the present disclosure is shown schematically.

[0020] Figure 2 A schematic diagram illustrating the principle of a satellite image damaged data reconstruction method according to an embodiment of the present disclosure is shown.

[0021] Figure 3 A schematic diagram illustrates the structural block diagram of a satellite image damaged data reconstruction apparatus based on a simulated annealing genetic algorithm according to an embodiment of the present disclosure;

[0022] Figure 4 A block diagram of an electronic device according to an embodiment of the present disclosure is shown schematically. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with specific embodiments and accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0024] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of a feature, step, operation, and / or component, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0025] In this disclosure, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this disclosure according to the specific circumstances.

[0026] In the description of this disclosure, it should be understood that the terms "longitudinal", "length", "circumferential", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this disclosure and simplifying the description, and do not indicate or imply that the subsystem or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this disclosure.

[0027] Throughout the accompanying drawings, identical elements are represented by the same or similar reference numerals. Conventional structures or configurations have been omitted where they may cause confusion in understanding this disclosure. Furthermore, the shapes, dimensions, and positional relationships of the components in the drawings do not reflect actual size, scale, or actual positional relationships. Additionally, any reference symbols enclosed in parentheses should not be construed as limiting.

[0028] Similarly, to simplify this disclosure and aid in understanding one or more of the various aspects of the disclosure, in the above description of exemplary embodiments of the present disclosure, various features of the present disclosure are sometimes grouped together in a single embodiment, figure, or description thereof. The use of terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refers to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present disclosure. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0029] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this disclosure, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0030] This disclosure provides a method for reconstructing damaged satellite image data, comprising: encoding an original image based on initialization parameters to obtain a first population consisting of multiple first individuals, each individual being a pre-reconstructed image corresponding to the original image, wherein the encoding method of the damaged area in the original image differs among different individuals; performing N iterations with an initial temperature as the starting temperature, where N is a positive integer; the iteration process includes: calculating the fitness of each first individual in the first population, the fitness being used to represent the closeness between the pre-reconstructed image and the original image; randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual; calculating the fitness of the second individual in the second population, and selecting the optimal individual in the second population based on the fitness; adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population; performing a cooling operation during each iteration, and in response to the current temperature meeting a preset condition, using the optimal individual in the second population as the reconstructed data of the image to be processed.

[0031] This embodiment of the disclosure dynamically switches between global search and local fine-tuning by adjusting crossover and mutation probabilities, combined with temperature control of simulated annealing. This avoids getting trapped in local optima and adds perturbations to the optimal individuals to escape the current local optimum and explore a better solution space. This achieves synergy between global optimization and local search, avoiding local optima. Through elite preservation and dynamic adjustment strategies, convergence performance and population diversity are effectively improved. By using fitness stretching and temperature control, image reconstruction accuracy and efficiency are balanced, maintaining stable convergence while possessing a strong ability to escape local optima.

[0032] Figure 1 A flowchart illustrating a method for reconstructing damaged satellite image data according to an embodiment of the present disclosure is shown. Figure 2 The schematic diagram illustrates the principle of a method for reconstructing damaged satellite image data according to an embodiment of the present disclosure.

[0033] like Figure 1 As shown, the satellite image damaged data reconstruction method of this embodiment includes operations S110 to S160.

[0034] In operation S110, the original image is encoded based on the initialization parameters to obtain a first population consisting of multiple first individuals. Each individual is a pre-reconstructed image corresponding to the original image, and the encoding method of the damaged area in the original image is different in different individuals.

[0035] In some embodiments, an initial population (i.e., a first population) can be generated based on the original image containing the damaged area and initialization parameters. For example, the initialization parameters may include initial temperature, cooling rate, termination temperature, maximum number of iterations N, and damaged area encoding method (such as pixel value padding, edge repair, frequency domain compensation, etc.).

[0036] M pre-reconstructed images I can be generated by applying different encoding strategies to the damaged areas of the original image. i (i.e., the first individual). Different pre-reconstructed images can be generated using different coding strategies, or they can be generated using different parameters under a unified coding strategy. The first group consists of multiple first individuals (pre-reconstructed images), each individual representing a possible restoration scheme.

[0037] In genetic algorithms, feasible solutions (i.e., possible restoration results) for an image can be described through encoding. For example, decimal encoding can be used to represent the value of each pixel in the image as a decimal number (i.e., gene p in a chromosome). i (i=1, 2, ..., n), concatenate the decimal values ​​of all pixels in the image in order to form a decimal string, resulting in a chromosome. Each chromosome can be represented by the geometric representation of multiple genes. In genetic algorithms, solutions to problems can be encoded as "chromosomes," where each decimal number in a chromosome corresponds to a pixel value that is a "gene" of that chromosome, allowing for operations such as crossover and mutation.

[0038] In some embodiments, N iterations can be performed using an initial temperature as the starting temperature. Each iteration process can include operations S120 to S150, where N is a positive integer.

[0039] In operation S120, the fitness of each first individual in the first group is calculated, and the fitness is used to represent the degree of proximity between the pre-reconstructed image and the original image.

[0040] In some embodiments, the fitness of each individual can be determined by calculating the closeness of each pre-reconstructed image to the original image. Fitness can be measured, for example, by mean squared error (MSE) or structural similarity (SSIM). MSE can be used to calculate the difference between corresponding pixels in the original and pre-reconstructed images; for example, the smaller the difference between the pixel values ​​of the target region in the pre-reconstructed image and the original image, the lower the MSE. SSIM focuses more on the similarity between brightness / contrast / structure (such as edge texture). Fitness can be used to measure the degree to which each individual in the population is likely to reach or approach the optimal solution in optimization computation.

[0041] In operation S130, two individuals from the first population are randomly selected for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual.

[0042] In some embodiments, two individuals can be randomly selected from the first population as parents, and a crossover operation can be performed based on the crossover probability. Offspring (i.e., second individuals) can be generated by exchanging some genes (i.e., pixel values) of the two parents. For example, the crossover probability can be dynamically adjusted based on fitness (e.g., individuals with high fitness have a low crossover probability, and individuals with low fitness have a high crossover probability). If an individual has a low crossover probability, the frequency of crossover based on that individual will be correspondingly lower.

[0043] For example, local regions of two pre-reconstructed images can be swapped using a single-point or double-point mapping method to form a new individual (i.e., a new reconstructed image).

[0044] In some embodiments, the mutation probability can be dynamically adjusted according to fitness. For example, individuals with high fitness have a low mutation probability and individuals with low fitness have a high mutation probability. By randomly modifying a gene in the offspring (such as adjusting the pixel value), a second individual and an offspring population containing the second individual (i.e., the second population) can be obtained.

[0045] Exemplarily, the mutation operation may include adjusting the pixels in a local area of the pre-reconstructed image, such as changing the color of the damaged area, etc.

[0046] In operation S140, calculate the fitness of the second individual in the second population, and screen out the optimal individual in the second population based on the fitness.

[0047] In some embodiments, after generating new individuals through selection, crossover, and mutation operations, calculate the fitness of the individuals in the second population, and screen out the optimal individual S0 in the second population. The optimal individual S0 may be the individual with the highest adaptability in the second population.

[0048] In operation S150, add random perturbations to the optimal individual, generate new individuals, and selectively put the new individuals back into the second population.

[0049] In some embodiments, add random perturbations to the optimal individual. For example, add tiny random noise (such as pixel value ±5%) to the optimal individual to generate new individuals. Calculate the fitness of the new individuals and selectively put the new individuals back based on the fitness of the new individuals.

[0050] Exemplarily, if the fitness of the new individual is better than that of the optimal individual, put the new individual back into the second population and replace the original optimal individual in the second population with the new individual. If the fitness of the new individual is lower than that of the optimal individual, receive the new individual based on a certain probability to avoid local optimal solutions.

[0051] In operation S160, perform a temperature reduction operation during each iteration process. In response to the current temperature satisfying a preset condition, determine the optimal individual in the second population as the reconstructed image of the image to be processed.

[0052] In some embodiments, perform a temperature reduction operation after each iteration to reduce the temperature. For example, the temperature T after each iteration is T = qt0, where t0 is the temperature in the previous iteration, q is the temperature reduction coefficient, and 0 < q < 1. Exemplarily, if the current iteration number is N, then T is the temperature of the Nth iteration, and t0 is the temperature in the (N - 1)th iteration. When the current temperature T is lower than the target temperature T end stop the iteration, and use the optimal individual in the current second population as the reconstructed image of the image to be processed. When the current temperature is not less than the target temperature T end continue to execute the iteration process until the preset condition is satisfied (that is, the current temperature T is lower than the target temperature T end )

[0053] The method disclosed herein dynamically adjusts crossover and mutation probabilities, combined with temperature control from simulated annealing, enabling the algorithm to dynamically switch between global search and local fine-tuning. This avoids getting trapped in local optima and adds perturbations to the optimal individuals to escape the current local optimum and explore a better solution space, achieving synergy between global optimization and local search, thus avoiding local optima. Through elite preservation and dynamic adjustment strategies, convergence performance and population diversity are effectively improved. By using fitness stretching and temperature control, image reconstruction accuracy and efficiency are balanced, maintaining stable convergence while possessing a strong ability to escape local extrema.

[0054] According to one embodiment of this disclosure, calculating the fitness of each first individual in a first group includes: determining the mean square error, structural similarity index, and image quality index of the original image and the pre-reconstructed image; determining the fitness of the first individual based on the mean square error, structural similarity index, and image quality index; and stretching the fitness of the first individual based on the temperature of the first group and the fitness of multiple first individuals in the first group to adjust the relative position of the first individual in the first group.

[0055] In some embodiments, the fitness of an individual can be represented by a fitness function f:

[0056]

[0057] in, , , , These are weighting coefficients used to adjust the contribution of different evaluation metrics to fitness. MSE is the mean squared error of pixel values ​​between the original and reconstructed images; fitness is related to... The fitness of two images is directly proportional to their relative properties; the smaller the MSE, the higher the fitness. SSIM is the structural similarity index, used to measure the structural similarity between two images, focusing particularly on the brightness, contrast pairs, and structural features of the images. The larger the SSIM, the higher the fitness. PSNR is used to measure image quality. The larger the PSNR, the higher the similarity between the reconstructed image and the original image. Penalty is a penalty term calculated based on certain specific conditions of the image, such as excessive smoothing or noise, used to reduce the fitness of unqualified solutions.

[0058] In some embodiments, the fitness of a single first individual is stretched based on the temperature of the first population and the fitness of multiple first individuals in the first population. The expression for the fitness stretching function f' is:

[0059]

[0060] Where f represents the fitness of each individual in the population, f maxrepresents the maximum fitness of an individual in the population, T is the current temperature of the population, and exp is the natural exponential function.

[0061] Fitness stretching maps the fitness of individuals in the population to a more reasonable range, resulting in a more uniform selection probability. This facilitates dynamic adaptation to temperature changes. After stretching, individuals with low fitness still have a certain probability of being selected, thus preserving potentially high-quality genes and preventing premature convergence. Fitness stretching effectively improves the robustness of the algorithm, weakens the impact of extreme values, and prevents the selection process from being dominated by a few individuals.

[0062] According to one embodiment of this disclosure, two first individuals are randomly selected from an initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individuals. This includes: classifying the first individuals based on preset conditions and the fitness of each first individual after stretching to obtain Class I individuals and Class II individuals; for Class I first individuals, directly copying the first individual to the second population; for Class II first individuals, randomly selecting two first individuals from the Class II individuals for adaptive crossover and adaptive mutation to obtain sub-individuals of the Class II individuals; adding at least some of the sub-individuals of the Class II individuals as second individuals to the second population; wherein, adaptive crossover is used to exchange pixel blocks in a certain pixel region between two first individuals; and adaptive mutation is used to randomly change a certain pixel block in the sub-individuals.

[0063] In some embodiments, the fitness probability P of each first individual in the first population can be calculated based on the stretched fitness. f And based on the fitness probability P f The first population is divided into two groups: Group 1 and Group 2. For Group 1 individuals, they are directly copied to the next generation (i.e., the second population) without crossover or mutation. For Group 2 individuals, adaptive crossover and adaptive mutation are performed. For example, the fitness probability P... f Individuals with fitness values ​​exceeding a specified value are classified as a single type and directly retained to the next generation. Fitness probability P f Individuals with values ​​less than or equal to the specified value are classified as binary individuals, and adaptive crossover and adaptive mutation are performed.

[0064] Wherein, fitness probability P f The expression is:

[0065]

[0066] In some embodiments, the adaptive crossover operation is used to perform pixel-level or block-level gene exchange between two parent individuals in two types of individuals to obtain child individuals. For example, the size or position of the crossover region can be dynamically determined based on image characteristics (such as texture complexity, distribution of damaged areas). For instance, large-block crossovers are used for smooth regions to accelerate convergence, while small-block crossovers are used for edge or textured regions to preserve details. The crossover operation may include exchanging pixel values ​​or image blocks in selected regions between two parent individuals. For example, the top-left 100×100 pixel region in two first individuals can be exchanged.

[0067] In some embodiments, adaptive mutation is performed on the sub-individuals obtained from the adaptive crossover operation to locally and randomly modify the offspring of the two types of individuals, introducing diversity. For example, a mutation region is selected within the sub-individuals, and adaptive mutation is achieved by randomly changing the pixel values ​​or parameters of the selected region. Mutation operations may include, for example, adding Gaussian noise to the mutation region, replacing the mutation region with the neighborhood mean, adjusting the contrast or brightness of the mutation region, etc.

[0068] At least a portion of the offspring (i.e., the second individuals) generated from the crossover and mutation of the two types of individuals are selected to join the second population. The second population includes directly replicated individuals of the first type and offspring obtained from the crossover and mutation of the second type of individuals. The size of the second population is the same as that of the initial population.

[0069] In some embodiments, the process of adaptive crossover and adaptive mutation further includes: calculating the crossover probability and mutation probability of each first individual based on the fitness statistics in the first population and the fitness of each first individual; the crossover probability is used to adjust the frequency of the crossover operation and the mutation probability is used to adjust the frequency of the mutation operation; and performing the crossover operation or mutation operation on the individual based on the crossover probability and mutation probability of the first individual.

[0070] For individuals of two types, the crossover probability P can be calculated using the fitness statistic of the first group and the fitness of the individual. c and the probability of mutation P m This allows the crossover and mutation probabilities to change with fitness.

[0071] When the individuals in a population exhibit significant differences and good diversity, the crossover and mutation probabilities decrease, preventing frequent crossovers from disrupting existing high-quality solutions or frequent mutations from introducing inferior genes (such as damaging repaired image regions). When the diversity of individuals in a population crosses over and their fitness tends to be consistent or to a local optimum, the crossover and mutation probabilities increase, effectively preventing the occurrence of local optima.

[0072] Adaptive crossover probability P c The expression can be:

[0073]

[0074] Among them, P c1 For the maximum crossover probability, P c2 For the minimum crossover probability, P c1 and P c2 f is a positive number less than 1. max f is the maximum fitness value in the current population. avg f is the average fitness value in the current population. m1 This represents the value with the higher fitness among the two crossed chromosomes.

[0075] Adaptive mutation probability P m The expression can be:

[0076]

[0077] Among them, P m1 P represents the maximum mutation probability. m2 For the minimum mutation probability, P m1 and P m2 f is a positive number less than 1, where f is the fitness value of an individual in the population.

[0078] The method provided in this disclosure dynamically adjusts the crossover and mutation probabilities based on the current population fitness state through adaptive crossover and mutation operations, balancing the algorithm's search capability and convergence speed, thereby improving the overall optimization performance.

[0079] According to one embodiment of this disclosure, adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population includes: randomly modifying the parameters of the optimal individual to obtain a new individual corresponding to the optimal individual; calculating the fitness of the new individual and comparing the fitness of the new individual with the fitness of the optimal individual; replacing the optimal individual in the second population with the new individual if the fitness of the new individual is not less than the fitness of the optimal individual; and accepting the new individual with a preset probability if the fitness of the new individual is less than the fitness of the optimal individual.

[0080] In some embodiments, the second-highest fit individual in the current population is selected as the optimal individual. The parameters of the optimal individual are then randomly modified, such as by randomly adding, subtracting, scaling, or replacing one or more parameters (e.g., RGB values, pixel values, etc. of an image). This random modification generates a "nearest solution" (i.e., a new individual) in the vicinity of the optimal individual. The new individual retains some high-quality features while introducing randomness to explore new possibilities.

[0081] For example, the fitness of the second population is calculated, the optimal individual S0 is selected, and S0 is replicated to obtain individual S1. S1 is then subjected to random perturbation (such as adjusting parameter values, modifying structure, etc.) to generate a new individual S. new Calculate the fitness of the new individual and compare it with the fitness of the optimal individual S0. If the fitness of the new individual is not less than that of the optimal individual, accept the new individual directly and replace the optimal individual with the new individual. At this time, the fitness of the new individual S0 is... new It is the best individual in the second group.

[0082] In some embodiments, the fitness values ​​of new individuals and the optimal individual after temperature adjustment can be calculated separately, and selective acceptance can be made based on the fitness values ​​after temperature adjustment. For the optimal individual S0, the fitness value C0 of this solution after temperature adjustment is calculated using the objective function; for the new individual S... new The fitness value C1 of the solution after temperature adjustment is calculated using the objective function.

[0083] The objective function is:

[0084]

[0085] in, It is a temperature-dependent adjustment factor that gradually increases the requirement for reconstruction accuracy as the temperature decreases; Let be the fitness function of an individual in the population.

[0086] New individual S new fitness f new If the fitness of the new individual S is not less than that of the optimal individual, then... new Replace the optimal individual S0. At this time, the new individual S new The current best individual is S. new fitness f new If the fitness of a new individual is less than that of the optimal individual (f0), a new individual is accepted with a certain probability (p). This process can be represented as:

[0087]

[0088] Where T is the current temperature.

[0089] The Metropolis criterion can be used to determine whether to accept a new solution as a loop, continuing until the number of loops exceeds the chain length L, at which point the loop ends. That is, if the fitness value of the new individual is greater than that of the original best individual, the new individual is used to replace the original best individual; if the fitness value of the new individual is less than that of the original best individual, the probability obtained from the Metropolis criterion is used to accept the new individual. This maintains population diversity and allows the genetic algorithm to converge quickly towards the optimal solution.

[0090] Figure 3 A schematic block diagram of a satellite image damaged data reconstruction apparatus based on a simulated annealing genetic algorithm according to an embodiment of the present disclosure is shown.

[0091] like Figure 3 As shown, the satellite image damaged data reconstruction device 300 based on simulated annealing genetic algorithm in this embodiment includes an encoding module 310, an iteration module 320 and a determination module 330.

[0092] The encoding module 310 is used to encode the original image based on initialization parameters to obtain a first population consisting of multiple first individuals, where each individual is a pre-reconstructed image corresponding to the original image, and the encoding method for the damaged areas in the original image differs among different individuals. In one embodiment, the encoding module 310 can be used to perform the operation S110 described above, which will not be repeated here.

[0093] The iteration module 320 is used to perform N iterations with an initial temperature as the starting temperature, where N is a positive integer. The iteration process includes: calculating the fitness of each first individual in the first population, where the fitness represents the closeness between the pre-reconstructed image and the original image; randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual; calculating the fitness of the second individual in the second population, and selecting the optimal individual in the second population based on the fitness; adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population. In one embodiment, the iteration module 320 can be used to perform operations S120 to S150 described above, which will not be repeated here.

[0094] The determination module 330 is used to perform a cooling operation during each iteration. In response to the current temperature meeting a preset condition, the optimal individual in the second population is determined as the reconstructed image of the image to be processed. In one embodiment, the determination and detection module 330 can be used to perform the operation S130 described above, which will not be repeated here.

[0095] According to embodiments of this disclosure, any plurality of modules among the encoding module 310, iteration module 320, and determination module 330 may be combined into one module, or any one of these modules may be split into multiple modules. Alternatively, at least a portion of the functionality of one or more of these modules may be combined with at least a portion of the functionality of other modules and implemented in one module. According to embodiments of this disclosure, at least one of the encoding module 310, iteration module 320, and determination module 330 may be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in any one of software, hardware, and firmware methods, or in a suitable combination of any of these. Alternatively, at least one of the encoding module 310, iteration module 320, and determination module 330 may be at least partially implemented as a computer program module, which, when run, can perform corresponding functions.

[0096] Figure 4 A block diagram of an electronic device according to an embodiment of the present disclosure is shown schematically.

[0097] like Figure 4 As shown, an electronic device 400 according to an embodiment of the present disclosure includes a processor 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage portion 408 into a random access memory (RAM) 403. The processor 401 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 401 may also include onboard memory for caching purposes. The processor 401 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0098] RAM 403 stores various programs and data required for the operation of electronic device 400. Processor 401, ROM 402, and RAM 403 are interconnected via bus 404. Processor 401 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 402 and / or RAM 403. It should be noted that the program may also be stored in one or more memories other than ROM 402 and RAM 403. Processor 401 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in one or more memories.

[0099] According to embodiments of this disclosure, the electronic device 400 may further include an input / output (I / O) interface 405, which is also connected to a bus 404. The electronic device 400 may also include one or more of the following components connected to the input / output (I / O) interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the input / output (I / O) interface 405 as needed. A removable medium 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 410 as needed so that computer programs read from it can be installed into the storage section 408 as needed.

[0100] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0101] According to embodiments of this disclosure, the computer-readable storage medium can be a non-volatile computer-readable storage medium. Examples include, but are not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0102] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Without departing from the scope of this disclosure, those skilled in the art can make various substitutions and modifications, all of which should be included within the protection scope of this disclosure.

Claims

1. A satellite image broken data reconstruction method based on an analog annealing genetic algorithm, characterized in that, The method includes: The original image is encoded based on the initialization parameters to obtain a first population consisting of multiple first individuals. Each individual is a pre-reconstructed image corresponding to the original image, and the encoding method of the damaged area in the original image is different in different individuals. The process is repeated N times, where N is a positive integer, starting from the initial temperature. The iterative process includes: Calculate the fitness of each individual in the first population, where the fitness represents the degree of closeness between the pre-reconstructed image and the original image; Two individuals are randomly selected from the initial population and subjected to adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual. Calculate the fitness of the second individual in the second population, and select the optimal individual in the second population based on the fitness. Add random perturbation to the optimal individual to generate a new individual and selectively return the new individual to the second population; A cooling operation is performed during each iteration. In response to the current temperature meeting a preset condition, the best individual in the second group is determined as the reconstructed image of the original image. The step of randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individuals includes: classifying the first individuals based on preset conditions and the fitness of each first individual after stretching to obtain a first-class individual and a second-class individual; for first individuals of class I, directly copying the first individual to the second population; for first individuals of class II, randomly selecting two first individuals from the second-class individuals for adaptive crossover and adaptive mutation to obtain sub-individuals of the second-class individuals; adding at least some sub-individuals of the second-class individuals as second individuals to the second population; the adaptive crossover is used to exchange pixel blocks in a certain pixel region between the two first individuals; the adaptive mutation is used to randomly change a certain pixel block in the sub-individual.

2. The method of claim 1, wherein, The calculation of the fitness of each individual in the first population includes: Perform the following operations on each of the first bodies: Determine the mean square error, structural similarity index, and image quality index of the original image and the pre-reconstructed image; The fitness of the first individual is determined based on the mean square error, structural similarity index, and image quality index. The fitness of the first individual is stretched based on the temperature of the first population and the fitness of multiple first individuals in the first population to adjust the relative position of the first individual in the first population.

3. The method of claim 1, wherein, The method further includes: Based on the fitness statistics in the first population and the fitness of each first individual, the crossover probability and mutation probability corresponding to each first individual are calculated; the crossover probability is used to adjust the frequency of the crossover operation, and the mutation probability is used to adjust the frequency of the mutation operation. Perform crossover or mutation operations on the individuals based on the crossover and mutation probabilities of the first individual.

4. The method of claim 1, wherein, The step of adding a random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population includes: Randomly modify the parameters of the optimal individual to obtain a new individual corresponding to the optimal individual; Calculate the fitness of the new individual and compare it with the fitness of the best individual; If the fitness of the new individual is not less than the fitness of the best individual, the best individual in the second population is replaced by the new individual. If the fitness of the new individual is less than that of the optimal individual, the new individual is accepted with a preset probability.

5. The method of claim 1, wherein, The method further includes: During the iteration process, a cooling operation is performed step by step. When the current temperature is lower than the target temperature, the best individual in the second group is used as the reconstructed image of the original image. If the current temperature is not lower than the target temperature, continue the iterative process.

6. A satellite image broken data reconstruction apparatus based on a simulated annealing genetic algorithm, characterized by, The device includes: The encoding module is used to encode the original image based on initialization parameters to obtain a first population consisting of multiple first individuals, wherein each individual is a pre-reconstructed image corresponding to the original image, and the encoding method of the damaged area in the original image is different in different individuals; An iterative module is used to perform N iterations with an initial temperature as the starting temperature, where N is a positive integer. The iterative process includes: calculating the fitness of each first individual in the first population, where the fitness represents the closeness between the pre-reconstructed image and the original image; randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individual; calculating the fitness of the second individual in the second population, and selecting the optimal individual in the second population based on the fitness; adding random perturbation to the optimal individual to generate a new individual and selectively returning the new individual to the second population. The determination module is used to perform a cooling operation during each iteration, and in response to the current temperature meeting a preset condition, determines the best individual in the second group as the reconstructed image of the original image; The step of randomly selecting two first individuals from the initial population for adaptive crossover and adaptive mutation to obtain a second individual and a second population containing the second individuals includes: classifying the first individuals based on preset conditions and the fitness of each first individual after stretching to obtain a first-class individual and a second-class individual; for first individuals of class I, directly copying the first individual to the second population; for first individuals of class II, randomly selecting two first individuals from the second-class individuals for adaptive crossover and adaptive mutation to obtain sub-individuals of the second-class individuals; adding at least some sub-individuals of the second-class individuals as second individuals to the second population; the adaptive crossover is used to exchange pixel blocks in a certain pixel region between the two first individuals; the adaptive mutation is used to randomly change a certain pixel block in the sub-individual.

7. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 5.

8. A computer readable storage medium having stored thereon a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 5.

9. A computer program product comprising computer programs or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 5.

Images