Open set processing method and device for unknown electromagnetic target

Abstract

The invention discloses an open set processing method and device for an unknown electromagnetic target. The method comprises the following steps: acquiring to-be-processed electromagnetic signal information; preprocessing the to-be-processed electromagnetic signal information to obtain target electromagnetic signal information; and performing identification processing on the target electromagnetic signal information to obtain target electromagnetic signal category information.

Description

Technical Field

[0001] The present invention relates to the field of electromagnetic technology, and in particular to an open set processing method and device for unknown electromagnetic targets. Background Art

[0002] Electromagnetic target perception and recognition is a key research area in the field of electromagnetic space security. With the emergence of artificial intelligence (AI) technology, intelligent and automatic perception and recognition of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves has become a growing trend. Existing intelligent recognition methods based on machine learning algorithms have effectively addressed the perception and recognition of known electromagnetic targets. However, when faced with unknown electromagnetic targets, because all target categories cannot be collected as training sets in a real electromagnetic environment, the model predicts all new categories as known ones, resulting in a loss of classification performance and subsequent errors. This loss and error is unacceptable in real-world applications. Therefore, a method and device for open-set processing of unknown electromagnetic targets is provided to extract invariant information between target and non-target categories, improve the accuracy of unknown electromagnetic target classification, and better distinguish known from unknown electromagnetic targets. This provides support for intelligent and automatic perception and recognition of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an open set processing method and device for unknown electromagnetic targets, which is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of the classification of unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0004] In order to solve the above technical problems, a first aspect of an embodiment of the present invention discloses a processing method, which includes:

[0005] Acquiring electromagnetic signal information to be processed;

[0006] Preprocessing the electromagnetic signal information to be processed to obtain target electromagnetic signal information;

[0007] The target electromagnetic signal information is identified and processed to obtain target electromagnetic signal category information.

[0008] A second aspect of an embodiment of the present invention discloses a processing device, comprising:

[0009] An acquisition module, used for acquiring electromagnetic signal information to be processed;

[0010] A first processing module is used to pre-process the electromagnetic signal information to be processed to obtain target electromagnetic signal information;

[0011] The second processing module is used to identify and process the target electromagnetic signal information to obtain target electromagnetic signal category information.

[0012] A third aspect of the present invention discloses another processing device, comprising:

[0013] a memory storing executable program code;

[0014] a processor coupled to the memory;

[0015] The processor calls the executable program code stored in the memory to execute part or all of the steps in the processing method disclosed in the first aspect of the embodiment of the present invention.

[0016] The fourth aspect of the present invention discloses a computer-readable storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the processing method disclosed in the first aspect of the embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 is a schematic diagram of a scenario of a processing system provided by an embodiment of the present invention;

[0019] Figure 2 It is a flowchart of a processing method disclosed in an embodiment of the present invention;

[0020] Figure 3 It is a structural diagram of a processing device disclosed in an embodiment of the present invention;

[0021] Figure 4 is a structural diagram of another processing device disclosed in an embodiment of the present invention;

[0022] Figure 5 Schematic diagram of the structure of a target open set recognition model disclosed in an embodiment of the present invention;

[0023] Figure 6 It is a structural diagram of a target noise reduction model disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0025] The terms "first," "second," and so on, in the description and claims of the present invention and the accompanying drawings are used to distinguish between different objects, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or device.

[0026] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0027] In this application, the word "exemplary" is used to mean "serving as an example, illustration, or illustration." Any embodiment described in this application as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is given to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that one of ordinary skill in the art can recognize that the present application can be implemented without using these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present application with unnecessary details. Therefore, the present application is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0028] It should be noted that since the method of the embodiment of the present application is executed in a computer device, the processing objects of each computer device exist in the form of data or information. For example, time is actually time information. It can be understood that if size, quantity, position, etc. are mentioned in subsequent embodiments, the corresponding data exist for the computer device to process. The details will not be repeated here.

[0029] It should be noted that the artificial intelligence related technologies that may be involved in this application are briefly described. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making.

[0030] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0031] Computer vision (CV) is the science of making machines "see." Specifically, it refers to machine vision, where cameras and computers replace the human eye in identifying and measuring objects, performing further image processing to create images more suitable for human observation or transmission to instruments. As a scientific discipline, computer vision studies related theories and technologies, attempting to build artificial intelligence systems capable of extracting information from images or multidimensional data. Computer vision technologies typically include image processing, image recognition, image semantic understanding, image retrieval, optical character recognition (OCR), video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous localization and mapping, and common biometric recognition technologies such as facial recognition and fingerprint recognition.

[0032] Unimodal information is data consisting of only one type, such as text, images, audio, video, or electromagnetic signals. Multimodal information is data that includes at least two types of unimodal information. Furthermore, multimodal information is suitable for complex tasks that require integrating multiple information sources, such as sentiment analysis, robot interaction, and autonomous driving. By integrating information from multiple modalities, higher performance and accuracy can often be achieved on the task.

[0033] A large model refers to an artificial neural network model with a very large number of parameters. In the field of artificial intelligence, a large model generally refers to a model with hundreds of millions to trillions of parameters. Models usually need to be trained on large-scale data sets and require a large amount of computing resources to be optimized and adjusted. Large models are generally used to solve complex tasks such as natural language processing, computer vision, and speech recognition. Generative AI is an AI that can create new content and ideas, including conversations, stories, images, videos, and music. In the embodiment of the present application, the large model can be ChatGPT, BERT, XLNet, Zhipu model, Claude, Moonshot AI model, ChatGLM model, Qianyi Tongwen model, MiniMax model, Spark model, Llama model, 360GPT model, Qwen model, Baichuan model, Skylark model, vivoLM model, Wenxin Yiyan and other large-scale language models, which are not limited in the embodiment of the present application.

[0034] The embodiments of the present application provide a processing method, an apparatus, a computer device, and a computer-readable storage medium, which are described in detail below.

[0035] See also Figure 1 , Figure 1 This is a schematic diagram of a processing system provided in an embodiment of the present application. The processing system may include a computer device 100, in which a processing device is integrated, such as Figure 1 Computer equipment in.

[0036] In the embodiment of the present application, the computer device 100 is mainly used to obtain electromagnetic signal information to be processed;

[0037] Preprocessing the electromagnetic signal information to be processed to obtain target electromagnetic signal information;

[0038] The target electromagnetic signal information is identified and processed to obtain the target electromagnetic signal category information.

[0039] It can extract the invariant information between target and non-target classes, improve the accuracy of classification of unknown electromagnetic targets, and thus better distinguish known electromagnetic targets from unknown electromagnetic targets, providing support for intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0040] In the embodiments of the present application, the computer device 100 may be an independent server, or a server network or server cluster composed of servers. For example, the computer device 100 described in the embodiments of the present application includes, but is not limited to, a computer, a network host, a single network server, a set of multiple network servers, or a cloud server composed of multiple servers. A cloud server is composed of a large number of computers or network servers based on cloud computing.

[0041] It is understood that the computer device 100 used in the embodiments of the present application can be a device that includes both receiving and transmitting hardware, that is, a device that has receiving and transmitting hardware capable of performing two-way communication over a two-way communication link. Such a device may include: a cellular or other communication device that has a single-line display, a multi-line display, or a cellular or other communication device without a multi-line display. The specific computer device 100 can be a desktop terminal or a mobile terminal. The computer device 100 can also be a mobile phone, a tablet computer, a laptop computer, etc.

[0042] Those skilled in the art will understand that Figure 1 The application environment shown in the figure is only one application scenario of the present application solution and does not constitute a limitation on the application scenario of the present application solution. Other application environments may also include Figure 1 More or fewer computer devices as shown in Figure 1 Only one computer device is shown in the figure. It can be understood that the processing system can also include one or more other services, which are not limited here.

[0043] In addition, if Figure 1 As shown, the processing system may further include a memory 200 for storing data, such as image data, location information, etc.

[0044] It should be noted that Figure 1 The scenario diagram of the processing system shown is only an example. The processing system and scenario described in the embodiment of the present application are intended to more clearly illustrate the technical solution of the embodiment of the present application, and do not constitute a limitation on the technical solution provided by the embodiment of the present application. Ordinary technicians in this field can know that with the evolution of the processing system and the emergence of new business scenarios, the technical solution provided by the embodiment of the present application is also applicable to similar technical problems.

[0045] The present invention discloses an open-set processing method and device for unknown electromagnetic targets. These methods facilitate the extraction of invariant information between target and non-target classes, improve the accuracy of unknown electromagnetic target classification, and further improve the distinction between known and unknown electromagnetic targets. This method supports the intelligent and automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves. These are described in detail below.

[0046] Example 1

[0047] See also Figure 2 , Figure 2 This is a flow chart of a processing method disclosed in an embodiment of the present invention. Figure 2 The processing method described is applied to a management system, such as a local server or cloud server for management, and is not limited in the embodiment of the present invention. Figure 2 As shown, the processing method may include the following operations:

[0048] 101. Obtain electromagnetic signal information to be processed.

[0049] 102. Preprocess the electromagnetic signal information to be processed to obtain target electromagnetic signal information.

[0050] 103. Identify and process the target electromagnetic signal information to obtain target electromagnetic signal category information.

[0051] It should be noted that conventional closed set classification is performed under known classification categories. Furthermore, the classic closed set training task is to use the training set Train the model and test it on the Tested on is a training signal, and y i ∈Y={1,2,…,K} is the corresponding label. In the closed set hypothesis, and From the same distribution The goal of the closed set algorithm is to fit a model f(x):X→Y that minimizes the expected risk:

[0052]

[0053] in is the hypothesis space, Is an indicator function that outputs 1 if the internal expression is true, otherwise it outputs 0. Assume that the model f is composed of two parts: feature embedding module and linear classifiers Right now The present invention defines the kth column of W as So W can be expressed as: W=[w1,…,w K ]. The output logit of the model on the kth class is: In general, the present invention uses the cross entropy loss function to optimize the model and maintain the model's discriminative ability on all known classes. To evaluate the performance of the model on the closed set, Also from the distribution The signal set extracted from .

[0054] In the open set training task scenario, the model is still based on the training set Training, however, the test set at this time Then it contains new types of signals, namely in is the corresponding label. The K+1th class here refers to all unseen classes, which may be more than one class. Since no other auxiliary information is included in the training set, the present invention cannot further subclassify all unseen classes. The optimal open set classifier should be able to minimize the following expected risk:

[0055]

[0056] because It is composed of known classes and unknown classes. Therefore, the expected risk in the above formula needs to classify the known classes at the same time and identify the unknown classes as K+1 classes. The classic machine learning algorithm outputs the class with the highest posterior probability, that is, However, since the model has not been exposed to new class signals during the training phase, its output probability for the K+1th class will always be the lowest value. Therefore, directly applying the closed set classifier to the open set recognition task will predict all new class signals as old class signals, resulting in a decrease in model performance. Therefore, a direct idea is to perform threshold division on the closed set classifier and record the maximum probability output by the model as the model's confidence in the signal, that is, The threshold method assumes that the model is more confident about the known category signal, and cannot give a high confidence prediction result for the unseen category. Therefore, the closed set classifier can be extended to the open set recognition task accordingly:

[0057]

[0058] Where th represents the set threshold. However, due to the overconfidence phenomenon of deep neural networks, the model's output confidence for both known and unknown categories is close to 1. Therefore, adjusting an appropriate threshold to distinguish between known and unknown categories becomes very difficult and time-consuming. Furthermore, because the semantic relationship between new and old categories varies across different open-set recognition tasks, the threshold setting in the above formula depends on this category-related semantic relationship.

[0059] It should be noted that the processing method of this application is adaptable to the classification of electromagnetic targets with small samples, which is more convenient and efficient to implement, and the embodiments of the present invention are not limited to this. This application verifies the effectiveness of the proposed method on a large number of data sets, including unknown detection tasks and open set recognition tasks, and learns adaptive threshold allocation for different category combinations, which the embodiments of the present invention are not limited to.

[0060] It should be noted that the above-mentioned electromagnetic signal information to be processed represents the received unknown electromagnetic pulse signal. Furthermore, through time-frequency analysis, time-frequency images representing different types of electromagnetic signals can be obtained, and then the electromagnetic signal information to be processed is preprocessed to obtain the target electromagnetic signal information. This embodiment of the present invention does not limit this.

[0061] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0062] In an optional embodiment, the target electromagnetic signal information is identified and processed to obtain target electromagnetic signal category information, including:

[0063] The target open set recognition model is used to perform feature extraction processing on the target electromagnetic signal information to obtain the target electromagnetic signal feature information;

[0064] The target electromagnetic signal characteristic information is identified and processed to obtain the target electromagnetic signal category information.

[0065] It should be noted that the loss function of the above target open set recognition model can be constructed based on the following formula:

[0066]

[0067] Where l can be cross entropy or other losses. The first term of the above formula is to match the true label of the signal, and its purpose is to maintain the classification performance of the classifier on the closed set signal. In the second term, This means setting the output of the model on the true mark to 0, is set to 0, and the second loss function matches the output probability of the model after being set to zero with the K+1th class, so that the output probability is the second largest among all classes. With the help of the above formula, the model can correctly classify the known category signal and make the output magnitude of the classifier between the target class and the non-target class. The embodiment of the present invention does not limit this.

[0068] Furthermore, the above-mentioned target open set recognition model can also be constructed based only on the L1 loss function, which can also meet the basic training requirements of the model of this application, and the embodiment of the present invention does not limit it.

[0069] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0070] In another optional embodiment, Figure 5 As shown, the target open set recognition model includes a first slicing module, a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a fifth convolution module, a sixth convolution module, a seventh convolution module, an eighth convolution module, a ninth convolution module, a tenth convolution module, a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a sixth feature extraction module, a seventh feature extraction module, an eighth feature extraction module, a first fusion module, a second fusion module, a third fusion module, a fourth fusion module, a fifth fusion module, a first sampling module, a second sampling module and a first pooling module; wherein,

[0071] The input end of the first slicing module is configured to receive the first model input of the target open set recognition model, and the output end of the first slicing module is connected to the input end of the first convolution module; the output end of the first convolution module is connected to the input end of the first feature extraction module; the output end of the first feature extraction module is connected to the input end of the second convolution module; the output end of the second convolution module is connected to the input end of the second feature extraction module; the output end of the second feature extraction module is connected to the input end of the third convolution module and the input end of the second fusion module respectively; the output end of the third convolution module is connected to the input end of the third feature extraction module; the output end of the third feature extraction module is connected to the input end of the fourth convolution module; the output end of the fourth convolution module is connected to the input end of the first fusion module and the input end of the first pooling module respectively; the output end of the first pooling module is connected to the input end of the fourth feature extraction module; the output end of the fourth feature extraction module is connected to the input end of the fifth convolution module; the output end of the fifth convolution module is connected to the input end of the fourth fusion module and the input end of the first sampling module respectively; the output end of the first sampling module is connected to the input end of the first fusion module; the output end of the first fusion module The output end is connected to the input end of the fifth feature extraction module; the output end of the fifth feature extraction module is connected to the input end of the sixth convolution module; the output end of the sixth convolution module is respectively connected to the input end of the third fusion module and the input end of the second sampling module; the output end of the second sampling module is connected to the input end of the second fusion module; the output end of the second fusion module is connected to the input end of the sixth feature extraction module; the output end of the sixth feature extraction module is respectively connected to the input end of the third fusion module and the input end of the tenth convolution module; the output end of the third fusion module is connected to the input end of the seventh feature extraction module; the output end of the seventh feature extraction module is connected to the input end of the seventh convolution module and the input end of the ninth convolution module; the output end of the seventh convolution module is connected to the input end of the fourth fusion module; the output end of the fourth fusion module is connected to the input end of the eighth feature extraction module; the output end of the eighth feature extraction module is connected to the input end of the eighth convolution module; the output end of the eighth convolution module, the output end of the ninth convolution module and the output end of the tenth convolution module are all connected to the input end of the fifth fusion module; the output end of the fifth fusion module is configured to output the first model output of the target open set recognition module.

[0072] It should be noted that the above-mentioned first slicing module is used to slice the input data information to reduce data loss during feature extraction and improve data processing speed. It can be constructed based on the Focus layer, which is not limited in the embodiment of the present invention.

[0073] It should be noted that the model architectures of the above-mentioned first feature extraction module, second feature extraction module, third feature extraction module, fourth feature extraction module, fifth feature extraction module, sixth feature extraction module, seventh feature extraction module, and eighth feature extraction module are consistent, and the embodiments of the present invention do not limit them. Furthermore, the above-mentioned first feature extraction module, second feature extraction module, third feature extraction module, fourth feature extraction module, fifth feature extraction module, sixth feature extraction module, seventh feature extraction module, and eighth feature extraction module are mainly used for fusing data features, which can achieve the goal of reducing network resource consumption while increasing the depth of feature fusion, thereby improving the processing speed, and the embodiments of the present invention do not limit them.

[0074] It should be noted that the convolution kernel size of the above-mentioned first convolution module, second convolution module, third convolution module, fourth convolution module, fifth convolution module, sixth convolution module, seventh convolution module, eighth convolution module, ninth convolution module and tenth convolution module can be one of 1×1×12, 1×1×32, 3×3×12, 3×3×32, 6×6×12 and 6×6×32, with a step size of 2, which is not limited in the embodiment of the present invention.

[0075] It should be noted that the above-mentioned first fusion module, second fusion module, third fusion module, fourth fusion module and fifth fusion module are constructed based on the Concat operation, which is not limited in the embodiment of the present invention.

[0076] It should be noted that the first sampling module and the second sampling module are constructed based on an upsampling operation (Upsample), which is not limited in the embodiment of the present invention.

[0077] It should be noted that the first pooling module described above is constructed based on the maximum pooling layer, which is not limited in this embodiment of the present invention. Furthermore, the first pooling module can be a maximum pooling layer, or it can be constructed by three parallel maximum pooling layers connected to a fusion module constructed based on the Concat operation, which is not limited in this embodiment of the present invention. Furthermore, the first pooling module can be one or more of 5×1×2, 9×1×4, and 13×1×6, which is not limited in this embodiment of the present invention.

[0078] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0079] In another optional embodiment, the first feature extraction module includes a first convolution unit, a second convolution unit, a third convolution unit, a fourth convolution unit, a fifth convolution unit, a first normalization unit, a second normalization unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a first activation unit, a second activation unit, a third activation unit, a fourth activation unit, a fifth activation unit, a first fusion unit and a second fusion unit; wherein,

[0080] The input end of the first convolution unit and the input end of the second convolution unit are configured as the input end of the first feature extraction module; the output end of the first convolution unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is connected to the input end of the first activation unit; the output end of the first activation unit is connected to the input end of the first fusion unit and the input end of the third convolution unit respectively; the output end of the third convolution unit is connected to the output end of the third normalization unit and the input end of the third activation unit; the output end of the third activation unit is connected to the input end of the fourth convolution unit; the output end of the fourth convolution unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is connected to the input end of the fourth activation unit. The output end of the fourth activation unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the second fusion unit; the output end of the second convolution unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the second activation unit; the output end of the second activation unit is connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the fifth convolution unit; the output end of the fifth convolution unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is connected to the input end of the fifth activation unit; the output end of the fifth activation unit is configured as the input end of the first feature extraction module.

[0081] It should be noted that the convolution kernel size of the above-mentioned first convolution unit, second convolution unit, third convolution unit, fourth convolution unit, and fifth convolution unit can be one of 1×1×12, 1×1×32, 3×3×12, 3×3×32, 6×6×12, and 6×6×32, with a step size of 2, which is not limited in this embodiment of the present invention.

[0082] It should be noted that the above-mentioned first normalization unit, second normalization unit, third normalization unit, fourth normalization unit, and fifth normalization unit are constructed based on the batch normalization layer, which is not limited in this embodiment of the present invention.

[0083] It should be noted that the above-mentioned first activation unit, second activation unit, third activation unit, fourth activation unit, and fifth activation unit can be constructed based on the Leaky Relu activation function or based on the SiLU activation function, which is not limited in the embodiment of the present invention.

[0084] It should be noted that the first fusion unit and the second fusion unit are constructed based on the Concat operation, which is not limited in the embodiment of the present invention.

[0085] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0086] In another optional embodiment, the target electromagnetic signal characteristic information is identified and processed to obtain target electromagnetic signal category information, including:

[0087] Obtaining category vector information; the category vector information includes several category vectors and category coefficients corresponding to the category vectors;

[0088] Using the first category calculation model to calculate and process the category vector information and the target electromagnetic signal characteristic information to obtain category calculation value information; the category calculation value information includes a plurality of category calculation values;

[0089] Among them, the first category calculation model is:

[0090] Kx i =‖TX-LB i ‖2;

[0091] Where Kx i Characterizes the calculated value of the i-th category in the calculated value information of the category; TX characterizes the characteristic information of the target electromagnetic signal; LB i Represents the i-th category vector in the category vector information;

[0092] The second category calculation model is used to calculate and process the category calculation value information and the category vector information to obtain coefficient calculation value information; the coefficient calculation value information includes a plurality of coefficient calculation values;

[0093] Among them, the second category calculation model is:

[0094]

[0095] In the formula, Ky i The calculated value of the coefficient in the coefficient calculation information; Kα i Represents the category coefficient corresponding to the i-th category vector in the category vector information;

[0096] Based on the target electromagnetic signal characteristic information and the coefficient calculation value information, the target electromagnetic signal category information is determined.

[0097] It should be noted that the above-mentioned category vector represents the category center vector corresponding to each electromagnetic target. It can be set by the user, or it can be a vector formed after clustering analysis of electromagnetic target signals through clustering algorithms such as K-means, or it can be obtained by fine-tuning the historical category center vector, which is not limited in this embodiment of the present invention. Furthermore, the category coefficient corresponding to the category vector is a positive number not less than 0, which can be set by the user, or it can be obtained by fine-tuning the historical category coefficient corresponding to the historical category center vector, which is not limited in this embodiment of the present invention. Furthermore, the category coefficient corresponding to the category vector can be a positive integer, which is not limited in this embodiment of the present invention.

[0098] In this optional embodiment, as an optional implementation manner, the above-mentioned determination of the target electromagnetic signal category information based on the target electromagnetic signal characteristic information and the coefficient calculation value information includes:

[0099] Performing product calculation processing on the target electromagnetic signal characteristic information and the coefficient calculation value information to obtain first product characteristic information; the first product characteristic information includes a plurality of first product characteristic vector information;

[0100] Using the third category calculation model to calculate and process the product feature information and the coefficient calculation value information to obtain second product feature vector information;

[0101] Among them, the third category calculation model is:

[0102]

[0103] Where Z represents the second product eigenvector information; KK i Represents the i-th first product feature vector information in the first product feature information; KL represents the number of coefficient calculation values;

[0104] Performing normalization activation processing on the first product feature vector information and the second product feature vector information in the first product feature information to obtain unit vector value information; the unit vector value information includes K+1 unit vector values;

[0105] Determine the maximum unit vector value in the unit vector value information as the target unit vector value;

[0106] The target category information corresponding to the target unit vector value is determined as the target electromagnetic signal category information.

[0107] It should be noted that the above determination of the target category information corresponding to the target unit vector value as the target electromagnetic signal category information is determined by analyzing the target category corresponding to each unit vector value. For example, when the correspondence between the unit vector value and the target category is as shown in the following table:

[0108] Target category Unit vector value Target Category A [0.9,1] Target Category B [0.8,0.9) Target Category C [0.6,0.8) Unknown category [0,0.6)

[0109] When the target unit vector value is 0.5, it is determined that the corresponding target category is an unknown category, which is not limited in this embodiment of the present invention.

[0110] It should be noted that the above-mentioned normalized activation processing of the first product feature vector information and the second product feature vector information in the first product feature information can be based on the softmax function or the LeakyReLU activation function, which is not limited in the embodiment of the present invention.

[0111] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0112] In an optional embodiment, the preprocessing of the electromagnetic signal information to be processed to obtain target electromagnetic signal information includes:

[0113] Performing conversion processing on the electromagnetic signal information to be processed to obtain electromagnetic conversion signal information;

[0114] The target noise reduction model is used to perform noise reduction processing on the electromagnetic conversion signal information to obtain the target electromagnetic signal information.

[0115] It should be noted that the above-mentioned conversion processing of the electromagnetic signal information to be processed is implemented based on the wavelet transform method, which is not limited by the embodiment of the present invention. Further, the CWD time-frequency image in the electromagnetic signal information to be processed is first subjected to wavelet decomposition to obtain the high-frequency component and low-frequency component corresponding to the CWD time-frequency image; at the same time, the WVD time-frequency image in the electromagnetic signal information to be processed is subjected to wavelet decomposition to obtain the high-frequency component and low-frequency component corresponding to the WVD time-frequency image; then the high-frequency component corresponding to the CWD time-frequency image and the high-frequency component corresponding to the WVD time-frequency image are fused (wavelet coefficients with similar amplitudes are fused based on the weighted average method, thereby retaining the image details common to the two types of time-frequency images, and wavelet coefficients with far different amplitudes are fused based on suppressing the smaller wavelet coefficients, The low-frequency components corresponding to the CWD time-frequency image and the low-frequency components corresponding to the WVD time-frequency image are fused (wavelet coefficients with similar amplitudes are fused based on the weighted average method to retain the original time-frequency characteristics of the signal, and wavelet coefficients with far different amplitudes are fused based on smaller wavelet coefficients to suppress redundant time-frequency characteristics in the CWD and WVD time-frequency images) to obtain the fused low-frequency components. The fused high-frequency components and the fused low-frequency components are then subjected to wavelet reconstruction to obtain electromagnetic conversion signal information representing the fused time-frequency image, which is not limited in the embodiment of the present invention.

[0116] It should be noted that the training of the target denoising model is based on a binary cross-entropy loss function, which is not limited in the present embodiment. The training set for training the target denoising model includes noisy samples and noise-free samples. The noisy samples serve as the training input data, and the noise-free samples serve as the supervisory signal, i.e., the input data label, to train the target denoising model, which is not limited in the present embodiment.

[0117] It should be noted that the above-mentioned target denoising model is a denoising model constructed based on a small convolution kernel, that is, a plurality of small-size convolution kernels are stacked to obtain the same receptive field as a large-size convolution kernel, which can reduce the noise of the time-frequency image represented by the electromagnetic conversion signal information, that is, the feature distribution difference of the time-frequency image is improved, thereby achieving its quality improvement under low signal-to-noise ratio conditions, which is not limited in the embodiments of the present invention.

[0118] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0119] In another optional embodiment, Figure 6As shown, the target denoising model includes an eleventh convolution module, a twelfth convolution module, a thirteenth convolution module, a fourteenth convolution module, a fifteenth convolution module, a first transposed convolution module, a second transposed convolution module, a third transposed convolution module, a fourth transposed convolution module, a fifth transposed convolution module, a sixth transposed convolution module, a seventh transposed convolution module, a second pooling module and a third pooling module; wherein,

[0120] The input end of the eleventh convolution module is configured to receive the second model input of the target denoising model; the output end of the eleventh convolution module is connected to the input end of the twelfth convolution module; the output end of the twelfth convolution module is connected to the input end of the second pooling module; the output end of the second pooling module is connected to the input end of the thirteenth convolution module; the output end of the thirteenth convolution module is connected to the input end of the fourteenth convolution module; the output end of the fourteenth convolution module is connected to the output end of the fifteenth convolution module, the input end of the fifteenth convolution module is connected to the input end of the third pooling module; the output end of the third pooling module is connected to the first transposed convolution module the input end of the first transposed convolution module; the output end of the second transposed convolution module is connected to the input end of the third transposed convolution module; the output end of the third transposed convolution module is connected to the input end of the fourth transposed convolution module; the output end of the fourth transposed convolution module is connected to the input end of the fifth transposed convolution module; the output end of the fifth transposed convolution module is connected to the input end of the sixth transposed convolution module; the output end of the sixth transposed convolution module is connected to the input end of the seventh transposed convolution module; the output end of the seventh transposed convolution module is configured to receive the second model output of the target denoising model.

[0121] It should be noted that the convolution kernel of the above-mentioned eleventh convolution module is 3×3×1, and the step size is 1; the convolution kernels of the twelfth convolution module and the thirteenth convolution module are 3×3×64, and the step size is 1; the convolution kernels of the fourteenth convolution module and the fifteenth convolution module are 3×3×128, and the step size is 1, which is not limited in the embodiment of the present invention.

[0122] It should be noted that the convolution kernel of the above-mentioned first transposed convolution module is 3×3×256, with a step size of 1; the convolution kernels of the second transposed convolution module and the third transposed convolution module are 3×3×128, with a step size of 2; the convolution kernel of the fourth transposed convolution module is 3×3×64, with a step size of 1; the convolution kernels of the fifth transposed convolution module and the sixth transposed convolution module are 3×3×32, with a step size of 1; the convolution kernel of the seventh transposed convolution module is 3×3×16, with a step size of 1, which is not limited in the embodiment of the present invention.

[0123] It should be noted that the above-mentioned second pooling module and third pooling module are constructed based on the maximum pooling layer, and their pooling kernels can be 2×2 with a step size of 2, which is not limited in the embodiment of the present invention.

[0124] Furthermore, an activation module may be configured after the seventh transposed convolution module, such as an activation layer constructed based on a sigmoid activation function or an activation layer constructed based on a leaky ReLU activation function, which is not limited in the embodiment of the present invention.

[0125] It can be seen that implementing the processing method described in the embodiment of the present invention is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, and providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0126] Example 2

[0127] See also Figure 3 , Figure 3 This is a schematic diagram of the structure of a processing device disclosed in an embodiment of the present invention. Figure 3 The described device can be applied to a management system, such as a local server or a cloud server for management, etc., and the embodiment of the present invention does not limit this. Figure 3 As shown, the device may include:

[0128] An acquisition module 201 is used to acquire electromagnetic signal information to be processed;

[0129] The first processing module 202 is used to pre-process the electromagnetic signal information to be processed to obtain target electromagnetic signal information;

[0130] The second processing module 203 is used to identify and process the target electromagnetic signal information to obtain target electromagnetic signal category information.

[0131] It can be seen that implementation Figure 3 The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0132] In another optional embodiment, Figure 3 As shown, the second processing module 203 identifies and processes the target electromagnetic signal information to obtain target electromagnetic signal category information, including:

[0133] The target open set recognition model is used to perform feature extraction processing on the target electromagnetic signal information to obtain the target electromagnetic signal feature information;

[0134] The target electromagnetic signal characteristic information is identified and processed to obtain the target electromagnetic signal category information.

[0135] It can be seen that implementation Figure 3 The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0136] In another optional embodiment, Figure 3 As shown, the target open set recognition model includes a first slicing module, a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a fifth convolution module, a sixth convolution module, a seventh convolution module, an eighth convolution module, a ninth convolution module, a tenth convolution module, a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a sixth feature extraction module, a seventh feature extraction module, an eighth feature extraction module, a first fusion module, a second fusion module, a third fusion module, a fourth fusion module, a fifth fusion module, a first sampling module, a second sampling module and a first pooling module; wherein,

[0137] The input end of the first slicing module is configured to receive the first model input of the target open set recognition model, and the output end of the first slicing module is connected to the input end of the first convolution module; the output end of the first convolution module is connected to the input end of the first feature extraction module; the output end of the first feature extraction module is connected to the input end of the second convolution module; the output end of the second convolution module is connected to the input end of the second feature extraction module; the output end of the second feature extraction module is connected to the input end of the third convolution module and the input end of the second fusion module respectively; the output end of the third convolution module is connected to the input end of the third feature extraction module; the output end of the third feature extraction module is connected to the input end of the fourth convolution module; the output end of the fourth convolution module is connected to the input end of the first fusion module and the input end of the first pooling module respectively; the output end of the first pooling module is connected to the input end of the fourth feature extraction module; the output end of the fourth feature extraction module is connected to the input end of the fifth convolution module; the output end of the fifth convolution module is connected to the input end of the fourth fusion module and the input end of the first sampling module respectively; the output end of the first sampling module is connected to the input end of the first fusion module; the output end of the first fusion module The output end is connected to the input end of the fifth feature extraction module; the output end of the fifth feature extraction module is connected to the input end of the sixth convolution module; the output end of the sixth convolution module is respectively connected to the input end of the third fusion module and the input end of the second sampling module; the output end of the second sampling module is connected to the input end of the second fusion module; the output end of the second fusion module is connected to the input end of the sixth feature extraction module; the output end of the sixth feature extraction module is respectively connected to the input end of the third fusion module and the input end of the tenth convolution module; the output end of the third fusion module is connected to the input end of the seventh feature extraction module; the output end of the seventh feature extraction module is connected to the input end of the seventh convolution module and the input end of the ninth convolution module; the output end of the seventh convolution module is connected to the input end of the fourth fusion module; the output end of the fourth fusion module is connected to the input end of the eighth feature extraction module; the output end of the eighth feature extraction module is connected to the input end of the eighth convolution module; the output end of the eighth convolution module, the output end of the ninth convolution module and the output end of the tenth convolution module are all connected to the input end of the fifth fusion module; the output end of the fifth fusion module is configured to output the first model output of the target open set recognition module.

[0138] It can be seen that implementation Figure 3 The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0139] In another optional embodiment, Figure 3As shown, the first feature extraction module includes a first convolution unit, a second convolution unit, a third convolution unit, a fourth convolution unit, a fifth convolution unit, a first normalization unit, a second normalization unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a first activation unit, a second activation unit, a third activation unit, a fourth activation unit, a fifth activation unit, a first fusion unit and a second fusion unit; wherein,

[0140] The input end of the first convolution unit and the input end of the second convolution unit are configured as the input end of the first feature extraction module; the output end of the first convolution unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is connected to the input end of the first activation unit; the output end of the first activation unit is connected to the input end of the first fusion unit and the input end of the third convolution unit respectively; the output end of the third convolution unit is connected to the output end of the third normalization unit and the input end of the third activation unit; the output end of the third activation unit is connected to the input end of the fourth convolution unit; the output end of the fourth convolution unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is connected to the input end of the fourth activation unit. The output end of the fourth activation unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the second fusion unit; the output end of the second convolution unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the second activation unit; the output end of the second activation unit is connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the fifth convolution unit; the output end of the fifth convolution unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is connected to the input end of the fifth activation unit; the output end of the fifth activation unit is configured as the input end of the first feature extraction module.

[0141] It can be seen that implementation Figure 3 The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0142] In another optional embodiment, Figure 3 As shown, the second processing module 203 identifies and processes the target electromagnetic signal characteristic information to obtain target electromagnetic signal category information, including:

[0143] Obtaining category vector information; the category vector information includes several category vectors and category coefficients corresponding to the category vectors;

[0144] Using the first category calculation model to calculate and process the category vector information and the target electromagnetic signal characteristic information to obtain category calculation value information; the category calculation value information includes a plurality of category calculation values;

[0145] Among them, the first category calculation model is:

[0146] Kx i =‖TX-LB i ‖2;

[0147] Where Kx i Characterizes the calculated value of the i-th category in the calculated value information of the category; TX characterizes the characteristic information of the target electromagnetic signal; LB i Represents the i-th category vector in the category vector information;

[0148] The second category calculation model is used to calculate and process the category calculation value information and the category vector information to obtain coefficient calculation value information; the coefficient calculation value information includes a plurality of coefficient calculation values;

[0149] Among them, the second category calculation model is:

[0150]

[0151] In the formula, Ky i The calculated value of the coefficient in the coefficient calculation information; Kα i Represents the category coefficient corresponding to the i-th category vector in the category vector information;

[0152] Based on the target electromagnetic signal characteristic information and the coefficient calculation value information, the target electromagnetic signal category information is determined.

[0153] It can be seen that implementation Figure 3 The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0154] In another optional embodiment, Figure 3 As shown, the first processing module 202 pre-processes the electromagnetic signal information to be processed to obtain target electromagnetic signal information, including:

[0155] Performing conversion processing on the electromagnetic signal information to be processed to obtain electromagnetic conversion signal information;

[0156] The target noise reduction model is used to perform noise reduction processing on the electromagnetic conversion signal information to obtain the target electromagnetic signal information.

[0157] It can be seen that implementation Figure 3The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0158] In another optional embodiment, Figure 3 As shown, the target denoising model includes an eleventh convolution module, a twelfth convolution module, a thirteenth convolution module, a fourteenth convolution module, a fifteenth convolution module, a first transposed convolution module, a second transposed convolution module, a third transposed convolution module, a fourth transposed convolution module, a fifth transposed convolution module, a sixth transposed convolution module, a seventh transposed convolution module, a second pooling module and a third pooling module; wherein,

[0159] The input end of the eleventh convolution module is configured to receive the second model input of the target denoising model; the output end of the eleventh convolution module is connected to the input end of the twelfth convolution module; the output end of the twelfth convolution module is connected to the input end of the second pooling module; the output end of the second pooling module is connected to the input end of the thirteenth convolution module; the output end of the thirteenth convolution module is connected to the input end of the fourteenth convolution module; the output end of the fourteenth convolution module is connected to the output end of the fifteenth convolution module, the input end of the fifteenth convolution module is connected to the input end of the third pooling module; the output end of the third pooling module is connected to the first transposed convolution module the input end of the first transposed convolution module; the output end of the second transposed convolution module is connected to the input end of the third transposed convolution module; the output end of the third transposed convolution module is connected to the input end of the fourth transposed convolution module; the output end of the fourth transposed convolution module is connected to the input end of the fifth transposed convolution module; the output end of the fifth transposed convolution module is connected to the input end of the sixth transposed convolution module; the output end of the sixth transposed convolution module is connected to the input end of the seventh transposed convolution module; the output end of the seventh transposed convolution module is configured to receive the second model output of the target denoising model.

[0160] It can be seen that implementation Figure 3 The processing device described is conducive to extracting the invariant information between the target class and the non-target class, improving the accuracy of classifying unknown electromagnetic targets, and thus better distinguishing known electromagnetic targets from unknown electromagnetic targets, providing support for the intelligent automatic perception and identification of electromagnetic targets based on electromagnetic information such as vortex electromagnetic waves and holographic electromagnetic waves.

[0161] Example 3

[0162] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of another processing device disclosed in an embodiment of the present invention. Figure 4The described device can be applied to a management system, such as a local server or a cloud server for management, etc., and the embodiment of the present invention does not limit this. Figure 4 As shown, the device may include:

[0163] A memory 301 storing executable program code;

[0164] a processor 302 coupled to the memory 301;

[0165] The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the processing method described in the first embodiment.

[0166] Example 4

[0167] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the processing method described in the first embodiment.

[0168] Example 5

[0169] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the processing method described in the first embodiment.

[0170] The device embodiments described above are merely illustrative. Modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, i.e., they may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0171] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus the necessary general hardware platform, or of course, by means of hardware. Based on this understanding, the above technical solution, in essence, or the portion that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0172] Finally, it should be noted that the open set processing method and device for unknown electromagnetic targets disclosed in the embodiments of the present invention are only preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A processing method, characterized in that: The method comprises: Acquiring electromagnetic signal information to be processed; Preprocessing the electromagnetic signal information to be processed to obtain target electromagnetic signal information; The target electromagnetic signal information is identified and processed to obtain target electromagnetic signal category information.

2. The processing method according to claim 1, characterized in that: The identifying and processing the target electromagnetic signal information to obtain the target electromagnetic signal category information includes: Using a target open set recognition model to perform feature extraction processing on the target electromagnetic signal information to obtain target electromagnetic signal feature information; The target electromagnetic signal characteristic information is identified and processed to obtain target electromagnetic signal category information.

3. The processing method according to claim 2, characterized in that: The target open set recognition model includes a first slicing module, a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a fifth convolution module, a sixth convolution module, a seventh convolution module, an eighth convolution module, a ninth convolution module, a tenth convolution module, a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a sixth feature extraction module, a seventh feature extraction module, an eighth feature extraction module, a first fusion module, a second fusion module, a third fusion module, a fourth fusion module, a fifth fusion module, a first sampling module, a second sampling module and a first pooling module; wherein, The input end of the first slicing module is configured to receive the first model input of the target open set recognition model, the output end of the first slicing module is connected to the input end of the first convolution module; the output end of the first convolution module is connected to the input end of the first feature extraction module; the output end of the first feature extraction module is connected to the input end of the second convolution module; the output end of the second convolution module is connected to the input end of the second feature extraction module; the output end of the second feature extraction module is connected to the input end of the third convolution module and the input end of the second fusion module respectively; the output end of the third convolution module is connected to the input end of the third feature extraction module; the output end of the third feature extraction module is connected to the input end of the fourth convolution module; the output end of the fourth convolution module is connected to the input end of the first fusion module and the input end of the first pooling module respectively; the output end of the first pooling module is connected to the input end of the fourth feature extraction module; the output end of the fourth feature extraction module is connected to the input end of the fifth convolution module; the output end of the fifth convolution module is connected to the input end of the fourth fusion module and the input end of the first sampling module respectively; the output end of the first sampling module is connected to the input end of the first fusion module; The output end is connected to the input end of the fifth feature extraction module; the output end of the fifth feature extraction module is connected to the input end of the sixth convolution module; the output end of the sixth convolution module is respectively connected to the input end of the third fusion module and the input end of the second sampling module; the output end of the second sampling module is connected to the input end of the second fusion module; the output end of the second fusion module is connected to the input end of the sixth feature extraction module; the output end of the sixth feature extraction module is respectively connected to the input end of the third fusion module and the input end of the tenth convolution module; the output end of the third fusion module is connected to the input end of the seventh feature extraction module; The output end of the seventh feature extraction module is connected to the input end of the seventh convolution module and the input end of the ninth convolution module; the output end of the seventh convolution module is connected to the input end of the fourth fusion module; the output end of the fourth fusion module is connected to the input end of the eighth feature extraction module; the output end of the eighth feature extraction module is connected to the input end of the eighth convolution module; the output end of the eighth convolution module, the output end of the ninth convolution module and the output end of the tenth convolution module are all connected to the input end of the fifth fusion module; the output end of the fifth fusion module is configured to output the first model output of the target open set recognition module.

4. The processing method according to claim 3, characterized in that: The first feature extraction module includes a first convolution unit, a second convolution unit, a third convolution unit, a fourth convolution unit, a fifth convolution unit, a first normalization unit, a second normalization unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a first activation unit, a second activation unit, a third activation unit, a fourth activation unit, a fifth activation unit, a first fusion unit and a second fusion unit; wherein, The input end of the first convolution unit and the input end of the second convolution unit are configured as the input end of the first feature extraction module; the output end of the first convolution unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is connected to the input end of the first activation unit; the output end of the first activation unit is respectively connected to the input end of the first fusion unit and the input end of the third convolution unit; the output end of the third convolution unit is connected to the output end of the third normalization unit and the input end of the third activation unit; the output end of the third activation unit is connected to the input end of the fourth convolution unit; the output end of the fourth convolution unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is connected to the fourth activation unit. The output end of the fourth activation unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the second fusion unit; the output end of the second convolution unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the second activation unit; the output end of the second activation unit is connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the fifth convolution unit; the output end of the fifth convolution unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is connected to the input end of the fifth activation unit; the output end of the fifth activation unit is configured as the input end of the first feature extraction module.

5. The processing method according to claim 2, characterized in that: The identifying and processing the target electromagnetic signal characteristic information to obtain the target electromagnetic signal category information includes: Acquire category vector information; the category vector information includes a plurality of category vectors and category coefficients corresponding to the category vectors; Using a first category calculation model to calculate and process the category vector information and the target electromagnetic signal characteristic information to obtain category calculation value information; the category calculation value information includes a plurality of category calculation values; Among them, the first category calculation model is: Kx i =‖TX-LB i ‖2; In the formula, Kx i represents the i-th category calculation value in the category calculation value information; TX represents the target electromagnetic signal characteristic information; LB i Representing the i-th category vector in the category vector information; Using a second category calculation model to calculate and process the category calculation value information and the category vector information to obtain coefficient calculation value information; the coefficient calculation value information includes a plurality of coefficient calculation values; Among them, the second category calculation model is: In the formula, Ky i Characterizes the i-th coefficient calculation value in the coefficient calculation value information; Kα i Representing a category coefficient corresponding to the i-th category vector in the category vector information; Based on the target electromagnetic signal characteristic information and the coefficient calculation value information, the target electromagnetic signal category information is determined.

6. The processing method according to claim 1, characterized in that: The preprocessing of the electromagnetic signal information to be processed to obtain target electromagnetic signal information includes: Converting the electromagnetic signal information to be processed to obtain electromagnetic conversion signal information; The target noise reduction model is used to perform noise reduction processing on the electromagnetic conversion signal information to obtain target electromagnetic signal information.

7. The processing method according to claim 6, characterized in that: The target denoising model includes an eleventh convolution module, a twelfth convolution module, a thirteenth convolution module, a fourteenth convolution module, a fifteenth convolution module, a first transposed convolution module, a second transposed convolution module, a third transposed convolution module, a fourth transposed convolution module, a fifth transposed convolution module, a sixth transposed convolution module, a seventh transposed convolution module, a second pooling module and a third pooling module; wherein, The input end of the eleventh convolution module is configured to receive the second model input of the target denoising model; the output end of the eleventh convolution module is connected to the input end of the twelfth convolution module; the output end of the twelfth convolution module is connected to the input end of the second pooling module; the output end of the second pooling module is connected to the input end of the thirteenth convolution module; the output end of the thirteenth convolution module is connected to the input end of the fourteenth convolution module; the output end of the fourteenth convolution module is connected to the output end of the fifteenth convolution module, the input end of the fifteenth convolution module is connected to the input end of the third pooling module; the output end of the third pooling module is connected to the first transposed convolution module The output end of the first transposed convolution module is connected to the input end of the second transposed convolution module; the output end of the second transposed convolution module is connected to the input end of the third transposed convolution module; the output end of the third transposed convolution module is connected to the input end of the fourth transposed convolution module; the output end of the fourth transposed convolution module is connected to the input end of the fifth transposed convolution module; the output end of the fifth transposed convolution module is connected to the input end of the sixth transposed convolution module; the output end of the sixth transposed convolution module is connected to the input end of the seventh transposed convolution module; the output end of the seventh transposed convolution module is configured to receive the second model output of the target denoising model.

8. A processing device, characterized in that: The device comprises: An acquisition module, used for acquiring electromagnetic signal information to be processed; A first processing module, used for preprocessing the electromagnetic signal information to be processed to obtain target electromagnetic signal information; The second processing module is used to identify and process the target electromagnetic signal information to obtain target electromagnetic signal category information.

9. A processing device, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the processing method according to any one of claims 1 to 7.

Images