CN117935237A - Method, system, equipment and storage medium for identifying space tiny targets - Google Patents
Method, system, equipment and storage medium for identifying space tiny targets Download PDFInfo
- Publication number
- CN117935237A CN117935237A CN202410108164.XA CN202410108164A CN117935237A CN 117935237 A CN117935237 A CN 117935237A CN 202410108164 A CN202410108164 A CN 202410108164A CN 117935237 A CN117935237 A CN 117935237A
- Authority
- CN
- China
- Prior art keywords
- solution set
- solution
- mutation
- space
- solutions
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 88
- 238000003860 storage Methods 0.000 title claims abstract description 11
- 230000035772 mutation Effects 0.000 claims abstract description 50
- 238000010367 cloning Methods 0.000 claims abstract description 32
- 230000008569 process Effects 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000004590 computer program Methods 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 4
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 29
- 238000012549 training Methods 0.000 description 15
- 238000010586 diagram Methods 0.000 description 13
- 230000006872 improvement Effects 0.000 description 6
- 230000004927 fusion Effects 0.000 description 4
- 230000035945 sensitivity Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 2
- 238000013095 identification testing Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a space tiny target identification method, a system, equipment and a storage medium, which belong to the technical field of data processing, wherein the method comprises the steps of extracting image characteristic data from a space target image in the rocket-based advancing process; mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution; performing mutation operation on the clone solution set to generate a mutation solution set, combining the mutation solution set with the original solution set to form a new solution set, deleting the new solution set, and updating the solution set to obtain a final solution set; and completing the identification of the space tiny target according to the image characteristics in the final solution set. The method can rapidly and accurately identify the event situation in the rocket advancing process, and is convenient for a rocket detection system to grasp the rocket flight state in real time.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method, a system, equipment and a storage medium for identifying a space tiny target.
Background
In the field of modern aerospace, one of the key links in assessing and verifying the flight phase of a carrier rocket is to observe the state of the rocket in the air. In the rocket flight process, the ground camera can shoot a rocket flight process diagram, and the rocket has the events such as interstage separation, multi-target separation and the like in the flight process. The traditional multi-target model needs to manually select an optimal model from the pareto optimal model set obtained through training to perform target recognition, and is not a completely automatic model. On the other hand, since many pareto optimal models exhibit similar performance, it is difficult to select an optimal model. In fact, many pareto optimal models are useful based on different small target recognition requirements, and the target recognition by manually selecting an optimal model is unreliable, and the flying state during rocket travel cannot be quickly recognized, so that the events during rocket travel cannot be quickly and accurately recognized.
Disclosure of Invention
Aiming at the problem that the prior art cannot quickly identify the flying state of the rocket in the advancing process, so that the event in the advancing process of the rocket cannot be quickly and accurately identified, the invention provides the space tiny target identification method, which can quickly and accurately identify the event in the advancing process of the rocket, and is convenient for a rocket detection system to master the flying state of the rocket in real time.
In order to achieve the above purpose, the invention adopts the following technical scheme:
A method for identifying a spatially tiny target, comprising:
Extracting image characteristic data from a space target image based on the rocket advancing process;
Mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
Performing mutation operation on the clone solution set to generate a mutation solution set, combining the mutation solution set with the original solution set to form a new solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And completing the identification of the space tiny target according to the image characteristics in the final solution set.
As a further improvement of the invention, the image characteristic data is extracted from the space target image based on the rocket advancing process, and then the image characteristic data is preprocessed to remove redundant characteristic data.
As a further improvement of the invention, the original solution set is cloned, and the generated solution forms a cloned solution set in the following specific way:
proportional cloning is used to maintain the best solution set, repeated multiple times to generate solutions with larger crowding distances,
Wherein: f i max is the maximum of the ith objective function; f i min is the minimum of the ith objective function; delta (D i, D) represents the crowding distance of the special solution D i;
Clone calculation was performed for δ (D i, D):
After performing a cloning operation on each solution set, all newly generated solutions constitute a clone solution set C (j).
As a further improvement of the invention, the clone solution set is subjected to mutation operation, the generated solution forms a mutation solution set, and the mutation solution set and the original solution set form a new solution set, and the concrete mode is as follows:
Performing mutation operation on the clone solution set C (j), wherein the mutation probability is MP, and generating a random mutation probability RP i for each individual in a special solution;
if MP > RP i, executing mutation, and after executing mutation operation on each solution, forming a mutation solution set M (j) by all new special solutions;
And combining the original solution set D (j) and the variant solution set M (j) to obtain a new solution set F (j).
As a further improvement of the invention, the final solution set is obtained after deleting and updating the new solution set, and the specific mode is as follows:
The same solutions exist in the new solution set F (j), so that the reduction of search space caused by the same solutions is avoided, deletion processing is performed on the same solutions, and only one of the same solutions is reserved;
If the number of solutions in the new solution set DF (j) after deleting the same solutions is smaller than the set value, carrying out cloning variation and deleting operation again until the number of solutions in the new solution set DF (j) is larger than the set value;
And if the number of solutions in the new solution set DF (j) after deleting the same solutions is larger than the set value, sequencing the solutions in the new solution set DF (j) after deleting the same solutions, and then selecting a plurality of solutions from the sequenced DF (j) to be made into a final solution set UD (j).
As a further improvement of the present invention, the output of the final solution set UD (j) needs to satisfy that the current iteration number is greater than the set maximum iteration number; if the current iteration number is greater than the set maximum iteration number, the identification of the space micro target is completed according to the image features in the final solution set.
As a further improvement of the present invention, the output of the final solution set UD (j) needs to satisfy that the current iteration number is greater than the set maximum iteration number; if the current iteration number is smaller than or equal to the set maximum iteration number, the final solution set UD (j) is taken as an initial solution set, and the solution of the final solution set is carried out again.
A spatially minute object recognition system comprising:
and an extraction module: the method comprises the steps of extracting image feature data from a space target image based on a rocket advancing process;
Cloning module: the method comprises the steps of mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
And deleting the updating module: the method comprises the steps of performing mutation operation on a clone solution set, forming a mutation solution set by the generated solution, forming a new solution set by the mutation solution set and an original solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And an identification module: the method is used for completing the identification of the space tiny targets according to the image features in the final solution set.
An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the one spatially minute object recognition method when the computer program is executed.
A computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the one space-micro target recognition method.
Compared with the prior art, the invention has the following beneficial effects:
The invention discloses a space tiny target recognition method, which is characterized in that a large number of image features capable of representing targets are extracted and analyzed from space target images for pareto model training, a new solution set is generated by cloning and mutation operation on a generated initial solution set, and a final solution set is obtained after deleting and updating the solution set on the new solution set; and finally, the identification of the space tiny target is completed through the image features in the final solution set. Training the extracted image features to obtain a pareto optimal model set in a training stage, calculating weights of all the pareto optimal models obtained in the training stage in a testing stage, inputting test sample features into each pareto optimal model with non-zero weights, and fusing the final output probability of the models to obtain the output of all the pareto optimal models through a evidence reasoning method. Compared with a method for manually selecting an optimal model, the method provided by the invention can be used for rapidly and accurately identifying the event in the rocket advancing process, is convenient for a rocket detection system to grasp the rocket flight state in real time, improves the speed and reliability of identifying the space tiny targets of the rocket and other aircrafts, and provides a reliable basis for the subsequent use planning of the rocket and other aircrafts.
Drawings
FIG. 1 is a schematic diagram of a method for identifying a space tiny object according to the present invention;
FIG. 2 is a schematic diagram of a method for identifying and testing a small space object according to the present invention;
FIG. 3 is a schematic diagram of a spatial target recognition framework for automatic multi-target learning in accordance with the present invention;
FIG. 4 is a schematic diagram of a comparison of a pareto optimal model with zero weight and non-zero weight;
FIG. 5 is a schematic diagram of a space micro target recognition system according to the present invention;
FIG. 6 is a schematic diagram of a system for identifying and testing a small target in space according to the present invention.
Detailed Description
The application will be described in detail below with reference to the drawings in connection with embodiments. It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other.
The following detailed description is exemplary and is intended to provide further details of the application. Unless defined otherwise, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the application.
Aiming at the problem that the flying state of the rocket in the advancing process cannot be rapidly identified, so that the event in the advancing process cannot be rapidly and accurately identified in the prior art, the invention provides a space tiny target identification method. As shown in fig. 1, the method includes:
Extracting image characteristic data from a space target image based on the rocket advancing process;
Mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
Performing mutation operation on the clone solution set to generate a mutation solution set, combining the mutation solution set with the original solution set to form a new solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And completing the identification of the space tiny target according to the image characteristics in the final solution set.
Based on the method for identifying the space tiny targets provided by the invention, the invention also provides a method for identifying and testing the space tiny targets, as shown in fig. 2, the method comprises the following steps:
Acquiring the weight of the pareto optimal model in the space tiny target identification method, and inputting the characteristics of the test sample into each pareto optimal model in a non-zero weight mode;
Fusing all pareto optimal models through an evidence reasoning method to obtain test output probability;
And obtaining a space target identification result according to the test output probability.
The invention is described in detail below with reference to the attached drawings:
As shown in fig. 3, it consists of two parts, a training phase and a testing phase. The training stage aims at generating the pareto optimal model set through multi-objective optimization. Because the weights are difficult to calculate during the training phase, the present invention calculates the weights for each model during the testing phase. And then inputting the characteristics of the test sample into the pareto optimal model obtained in the training stage by non-zero weight. The final output probabilities of the models are obtained by fusing the output probabilities of the models using a evidence reasoning method. Finally, the label with the maximum output probability is regarded as a final label. The scheme is developed according to the following flow:
S1: feature extraction and data preprocessing
A number of image features characterizing the target, including intensity features, texture features, and geometric features, are extracted and analyzed from the spatial target image for pareto model training. And preprocessing the data after extracting the features, removing redundant features, and reducing the calculated amount.
S2: in order to avoid the influence of feature selection on pareto model training, feature selection and model parameter training are performed simultaneously. Let the model parameters be defined as α= { α 1,…,αM }, M be the number of model parameters. All features are defined as β= { β 1,…,βN }, N being the number of features. The optimization objective of the model is to maximize sensitivity (f sen) and specificity (f spe) simultaneously to obtain the pareto optimal model set, as shown in equation (1).
f＝maxα,β(fsen,fspe) (1)
Wherein: f is an objective function; f sen is sensitivity; f spe is specificity.
S3: initializing: because feature selection and pareto model parameter training are performed simultaneously, a hybrid initialization of the selected features and pareto model parameters is required, and an initial solution set is randomly generated. A solution consists of a set of binary or integer values, called an "individual set". Each individual in a solution represents a feature or a model parameter. The features are encoded by a binary encoding method, with a value of '1' indicating that the corresponding feature is selected and '0' indicating that the corresponding feature is not selected. Because the values of the pareto model parameters are continuous, the pareto model parameters are directly optimized. Let G max denote the maximum number of iterations, D (i) = { D 1,…,dp } be defined as a solution set, where D i, i=1, …, p is the special solution.
S4: cloning
The invention uses proportional cloning to maintain the best solution set, repeatedly generates solutions with larger crowding distances for a plurality of times, delta (D i, D) represents the crowding distance of special solutions D i, and the calculation method is shown as the formula (2)
Wherein: f i max is the maximum of the ith objective function; f i min is the minimum of the ith objective function;
The calculation method of delta i (D, D) is shown in the formula (3).
After performing a cloning operation on each solution set, all newly generated solutions constitute a clone solution set, denoted by C (j).
S5: variation of
In order to obtain better effect, mutation operation is carried out on the cloned solution set C (j). Assuming that the variation probability is defined as MP, a random variation probability RP i is generated for each individual in a solution. If MP > RP i, the mutation is performed. After performing the mutation operation on each solution, all new special solutions form a mutation solution set M (j). The original solution sets D (j) and M (j) are combined to obtain a new solution set F (j).
S6: deletion of
After performing the two steps, the same solution may exist in the newly generated solution set F (j). To avoid the same solution causing a reduction in search space, only a unique solution is reserved. All the remaining solutions constitute a new solution set DF (j). If the number of solutions in the solution set DF (j) is smaller than P, the process returns to step S4, otherwise step S7 is performed.
S7: updating solution sets
To guarantee solution set size, P special solutions need to be selected from DF (j). Before performing the update operation, the present invention evaluates the performance metrics of each ad hoc solution, and thus in this step ranks the solutions sets DF (j) in descending order according to the AUC of each solution using a fast non-dominant ranking method, and then selects the first P solutions from the ranked DF (j) to form a new solution set UD (j).
S8: if Z > G max, UD (j) is output, the algorithm ends, otherwise z=z+1, d (j) =ud (j), repeating S4.
Wherein Z is the current iteration number.
Unlike model training based on a single objective, a pareto optimal model set f= [ F 1,…,FJ ] is obtained after training, consisting of a number of models with different model parameters and feature combinations, as shown in fig. 3, where J is the number of pareto optimal models.
S9: the testing phase uses all models in the pareto optimal model set to increase the diversity of the models and thereby improve the performance of the testing phase. However, the present invention also contemplates the relative weights of the models, since the contribution of different pareto optimal models to the prediction results may be different. The probability outputs of all pareto optimal models are combined together through an effective fusion method so as to improve the performance of the models and make the models more stable. Thus, the test phase consists of two parts, weight calculation and evidence-based reasoning fusion. For each model F j (j=1, … J) in the pareto optimal model set, the weight calculation method is shown in formula (4).
Wherein: to test the sensitivity of the sample to model F j; /(I) To test the specificity of the sample for model F j; AUC j is the AUC of the test sample for model F j; λ is a weight vector 0< λ <1 that adjusts the importance of two parts in w j.
Since a balanced result is desired, the weights of the models that achieve a good balance between sensitivity and specificity are non-zero, while the weights of the other models are zero. Fig. 4 shows a pareto optimal model with two types of weights. On the other hand, since AUC is often used to measure reliability of a model, it is also considered in weight calculation. Although the present invention uses all pareto optimal models, the model with zero weight has no effect on the final result. After the weights are obtained, they are normalized using equation (5).
The invention finally merges a plurality of pareto optimal models with non-zero weights. For each of the test specimens to be tested,Respectively, the probability that model F j predicts it as the kth class, and/>The final output probability P k is obtained through evidence reasoning fusion, and the fusion method is shown in a formula (6).
Wherein:
the sample final label L is determined by equation (7).
L＝max(Pk) (7)
Wherein: p k is the probability output for each category after fusion.
As shown in fig. 5, a third object of the present invention is to provide a space-based micro target recognition system, comprising:
and an extraction module: the method comprises the steps of extracting image feature data from a space target image based on a rocket advancing process;
Cloning module: the method comprises the steps of mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
And deleting the updating module: the method comprises the steps of performing mutation operation on a clone solution set, forming a mutation solution set by the generated solution, forming a new solution set by the mutation solution set and an original solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And an identification module: the method is used for completing the identification of the space tiny targets according to the image features in the final solution set.
A fourth object of the present invention is to provide an electronic device comprising a memory, a processor and a computer program stored in said memory and executable on said processor, said processor implementing the steps of said method for identifying spatially small objects when said computer program is executed.
The method for identifying the space tiny targets comprises the following steps:
Extracting image characteristic data from a space target image based on the rocket advancing process;
Mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
Performing mutation operation on the clone solution set to generate a mutation solution set, combining the mutation solution set with the original solution set to form a new solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And completing the identification of the space tiny target according to the image characteristics in the final solution set.
A fifth object of the present invention is to provide a computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the space-micro target recognition method.
The method for identifying the space tiny targets comprises the following steps:
Extracting image characteristic data from a space target image based on the rocket advancing process;
Mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
Performing mutation operation on the clone solution set to generate a mutation solution set, combining the mutation solution set with the original solution set to form a new solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And completing the identification of the space tiny target according to the image characteristics in the final solution set.
As shown in fig. 6, a sixth object of the present invention is to provide a space micro target recognition test system, comprising:
And (3) an identification input module: the method is used for acquiring the weight of the pareto optimal model in the space tiny target identification method, and inputting the characteristics of the test sample into each pareto optimal model in a non-zero weight mode;
and a fusion module: the method is used for fusing all pareto optimal models through a evidence reasoning method to obtain test output probability;
And an identification result module: and the method is used for obtaining a space target recognition result according to the test output probability.
A seventh object of the present invention is to provide an electronic device comprising a memory, a processor and a computer program stored in said memory and executable on said processor, said processor implementing the steps of said one method for spatially minute object recognition testing when executing said computer program.
The space tiny target identification test method comprises the following steps:
Acquiring the weight of the pareto optimal model in the space tiny target identification method, and inputting the characteristics of the test sample into each pareto optimal model in a non-zero weight mode;
Fusing all pareto optimal models through an evidence reasoning method to obtain test output probability;
And obtaining a space target identification result according to the test output probability.
An eighth object of the present invention is to provide a computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the space micro target recognition test method.
The space tiny target identification test method comprises the following steps:
Acquiring the weight of the pareto optimal model in the space tiny target identification method, and inputting the characteristics of the test sample into each pareto optimal model in a non-zero weight mode;
Fusing all pareto optimal models through an evidence reasoning method to obtain test output probability;
And obtaining a space target identification result according to the test output probability.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
Claims (10)
1. A method for identifying a spatially minute object, comprising:
Extracting image characteristic data from a space target image based on the rocket advancing process;
Mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
Performing mutation operation on the clone solution set to generate a mutation solution set, combining the mutation solution set with the original solution set to form a new solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And completing the identification of the space tiny target according to the image characteristics in the final solution set.
2. The method for identifying a space tiny target according to claim 1, wherein the image feature data is extracted from a space target image based on a rocket advancing process, and then the image feature data is preprocessed to remove redundant feature data.
3. The method for identifying a small space target according to claim 1, wherein the cloning of the original solution set is performed in the following specific manner that the generated solution forms a cloned solution set:
proportional cloning is used to maintain the best solution set, repeated multiple times to generate solutions with larger crowding distances,
Wherein: f i max is the maximum of the ith objective function; f i min is the minimum of the ith objective function; delta (D i, D) represents the crowding distance of the special solution D i;
Clone calculation was performed for δ (D i, D):
After performing a cloning operation on each solution set, all newly generated solutions constitute a clone solution set C (j).
4. The method for identifying a small space object according to claim 1, wherein the mutation operation is performed on the clone solution set, and the generated solution forms a mutation solution set, wherein the mutation solution set and the original solution set form a new solution set by the following specific steps:
Performing mutation operation on the clone solution set C (j), wherein the mutation probability is MP, and generating a random mutation probability RP i for each individual in a special solution;
if MP > RP i, executing mutation, and after executing mutation operation on each solution, forming a mutation solution set M (j) by all new special solutions;
And combining the original solution set D (j) and the variant solution set M (j) to obtain a new solution set F (j).
5. The method for identifying a tiny space object according to claim 1, wherein the final solution set is obtained after deleting and updating the new solution set, and the specific manner is as follows:
The same solutions exist in the new solution set F (j), so that the reduction of search space caused by the same solutions is avoided, deletion processing is performed on the same solutions, and only one of the same solutions is reserved;
If the number of solutions in the new solution set DF (j) after deleting the same solutions is smaller than the set value, carrying out cloning variation and deleting operation again until the number of solutions in the new solution set DF (j) is larger than the set value;
And if the number of solutions in the new solution set DF (j) after deleting the same solutions is larger than the set value, sequencing the solutions in the new solution set DF (j) after deleting the same solutions, and then selecting a plurality of solutions from the sequenced DF (j) to be made into a final solution set UD (j).
6. The method for identifying a small space object according to claim 5, wherein the output of the final solution set UD (j) needs to satisfy that the current iteration number is greater than the set maximum iteration number;
if the current iteration number is greater than the set maximum iteration number, the identification of the space micro target is completed according to the image features in the final solution set.
7. The method for identifying a small space object according to claim 5, wherein the output of the final solution set UD (j) needs to satisfy that the current iteration number is greater than the set maximum iteration number;
if the current iteration number is smaller than or equal to the set maximum iteration number, the final solution set UD (j) is taken as an initial solution set, and the solution of the final solution set is carried out again.
8. A spatially minute object recognition system, comprising:
and an extraction module: the method comprises the steps of extracting image feature data from a space target image based on a rocket advancing process;
Cloning module: the method comprises the steps of mixing and initializing image characteristic data and pareto model parameters, randomly generating an initial solution set, cloning the initial solution set, and forming a cloning solution set by the generated solution;
And deleting the updating module: the method comprises the steps of performing mutation operation on a clone solution set, forming a mutation solution set by the generated solution, forming a new solution set by the mutation solution set and an original solution set, deleting the new solution set, and updating the solution set to obtain a final solution set;
And an identification module: the method is used for completing the identification of the space tiny targets according to the image features in the final solution set.
9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of a method of spatial micro-object identification according to any one of claims 1-7 when the computer program is executed.
10. A computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of a method of spatial micro-object identification according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410108164.XA CN117935237A (en) | 2024-01-25 | 2024-01-25 | Method, system, equipment and storage medium for identifying space tiny targets |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410108164.XA CN117935237A (en) | 2024-01-25 | 2024-01-25 | Method, system, equipment and storage medium for identifying space tiny targets |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117935237A true CN117935237A (en) | 2024-04-26 |
Family
ID=90757147
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410108164.XA Pending CN117935237A (en) | 2024-01-25 | 2024-01-25 | Method, system, equipment and storage medium for identifying space tiny targets |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117935237A (en) |
-
2024
- 2024-01-25 CN CN202410108164.XA patent/CN117935237A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113657465B (en) | Pre-training model generation method and device, electronic equipment and storage medium | |
| CN109461001B (en) | Method and device for obtaining training sample of first model based on second model | |
| US20170091675A1 (en) | Production equipment including machine learning system and assembly and test unit | |
| CN105095494B (en) | The method that a kind of pair of categorized data set is tested | |
| CN111428849A (en) | Improved particle swarm algorithm-based transfer function model parameter identification method and device | |
| CN107832789B (en) | Feature weighting K nearest neighbor fault diagnosis method based on average influence value data transformation | |
| CN112180853A (en) | Flexible job shop scheduling hybrid optimization method based on multi-population strategy | |
| CN110941902A (en) | Lightning stroke fault early warning method and system for power transmission line | |
| CN110222824B (en) | Intelligent algorithm model autonomous generation and evolution method, system and device | |
| CN109731338A (en) | Artificial intelligence training method and device, storage medium and electronic device in game | |
| CN117935237A (en) | Method, system, equipment and storage medium for identifying space tiny targets | |
| CN108549899A (en) | A kind of image-recognizing method and device | |
| CN110929851A (en) | AI model automatic generation method based on computational graph subgraph | |
| CN115345303A (en) | Convolutional neural network weight tuning method, device, storage medium and electronic equipment | |
| CN110070104A (en) | A kind of hyper parameter determines method, apparatus and server | |
| CN109756494B (en) | Negative sample transformation method and device | |
| Park et al. | Niching genetic programming based hyper-heuristic approach to dynamic job shop scheduling: an investigation into distance metrics | |
| US11514268B2 (en) | Method for the safe training of a dynamic model | |
| CN113657604A (en) | Device and method for operating an inspection table | |
| Yan et al. | An improved imbalanced data classification algorithm based on SVM | |
| CN117670095B (en) | Method and device for generating action plan of multi-agent autonomous countermeasure drive | |
| Park et al. | Enhancing heuristics for order acceptance and scheduling using genetic programming | |
| WO2023214582A1 (en) | Learning device, learning method, and learning program | |
| CN111931416B (en) | Hyper-parameter optimization method for graph representation learning combined with interpretability | |
| CN116520281B (en) | DDPG-based extended target tracking optimization method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |