SAR main shaft prediction data set construction method based on TransCGAN simulation parameter extraction framework

Abstract

The invention provides an SAR main shaft prediction data set construction method based on a TransCGAN simulation parameter extraction framework, and the method comprises the steps: obtaining a target 3D model, and obtaining a simulation SAR image data set according to the target 3D model; a parameter prediction network based on a TransCGAN model is constructed; acquiring a real SAR image, and inputting the simulation SAR image data set and the real SAR image into a parameter prediction network to obtain predicted target to-be-measured parameters; and obtaining a target SAR main shaft prediction data set according to the predicted target to-be-measured parameters. According to the method, a target 3D model and a real SAR image serve as input data, a TransCGAN network is trained to obtain the relation between the SAR image and simulation parameters, conversion from the simulation parameters to the simulation SAR image is achieved with the help of software, and finally accurate simulation parameters are extracted to obtain a target SAR main axis prediction data set which is used for training a target main axis direction prediction network.

Description

Technical Field

[0001] The present invention relates to the technical field of synthetic aperture radar target main axis direction prediction, and in particular to a SAR main axis prediction data set construction method based on a TransCGAN simulation parameter extraction framework. Background Art

[0002] Synthetic Aperture Radar (SAR) has the ability to work around the clock and in all weather conditions. It is an advanced remote sensing technology that uses microwave signals to monitor the surface of the earth. It captures surface information by actively emitting microwave signals and receiving their reflected signals. High-resolution SAR images can provide rich details of the target, such as structure, shape, material properties, etc.

[0003] SAR image interpretation is very important in both military and civilian fields, especially in target detection and recognition. In the process of SAR image interpretation, determining the main axis direction of the target is crucial for tasks such as target recognition, attitude estimation and tracking. The main axis direction of the target can be described by azimuth. Azimuth angle, also known as horizontal longitude (Azimuth angle, abbreviated as Az), is one of the methods for measuring the angle of an object on a plane.

[0004] In the field of machine learning, building a high-quality dataset is the key to successfully training a model. The main axis prediction dataset contains a large number of SAR image datasets labeled with the main axis direction of the target, which can be used to train and optimize the model. However, SAR images of the same target under different imaging conditions are often scarce, and data acquisition is difficult. In order to overcome this challenge, SAR image simulation technology came into being, which can quickly provide SAR target simulation images under different scenes and imaging conditions at a low cost, thereby supplementing the lack of actual data.

[0005] The key to SAR image simulation lies in the accuracy of simulation parameters, which are crucial for generating high-quality simulation images. Traditional simulation parameter acquisition methods, including field measurements and manual adjustments, are not only time-consuming and labor-intensive, inefficient, but also difficult to ensure the accuracy of parameters. In addition, the relationship between SAR images and simulation parameters is complex and difficult to describe through explicit functional relationships, which further increases the difficulty of extracting simulation parameters from real SAR images. Summary of the invention

[0006] Based on this, it is necessary to provide a method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework to address the above technical problems.

[0007] The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework includes the following steps:

[0008] Acquire a target 3D model, and obtain a simulated SAR image data set according to the target 3D model;

[0009] Construct a parameter prediction network based on the TransCGAN model;

[0010] Acquire a real SAR image data set, input the simulated SAR image and the real SAR image into the parameter prediction network, and obtain predicted target parameters to be measured;

[0011] According to the predicted target parameters to be measured, a target SAR principal axis prediction data set is obtained.

[0012] In one embodiment, acquiring a target 3D model, and obtaining a simulated SAR image data set according to the target 3D model includes:

[0013] Preprocessing the target 3D model to obtain a processed 3D model;

[0014] The processed 3D model is simulated by image simulation software to obtain a simulated SAR image data set.

[0015] In one embodiment, simulating the processed 3D model through image simulation software to obtain a simulated SAR image data set includes:

[0016] Obtain the value range of specific simulation parameters and target parameters to be measured;

[0017] Inputting the specific simulation parameters and the processed 3D model into the image simulation software, sampling and taking values ​​of the target parameters to be measured one by one within the value range, so as to perform simulation and obtain a plurality of simulated SAR images;

[0018] The multiple simulated SAR images constitute the simulated SAR image dataset.

[0019] In one embodiment, constructing a parameter prediction network based on the TransCGAN model includes:

[0020] The parameter prediction network is shown below:

[0021]

[0022] Where L represents the loss function, G represents the generator, D represents the discriminator, min represents the minimum value, max represents the maximum value, V(D,G) represents the parameter prediction network, E[·] represents the expected function, x represents the real data, i.e., the real SAR image; y represents the conditional information, i.e., the simulated SAR image, and p z(z) represents the distribution of input noise, z represents the input noise of the generator, G(z,y) represents the predicted simulation parameters generated by the generator; D(x,y), D(G(z,y),y) represent the discrimination results of the discriminator, p data (x) represents the true distribution of samples in the data set, and log represents the logarithmic operation.

[0023] In one embodiment, it also includes:

[0024] The generator is a Swin Transformer network based on the attention mechanism;

[0025] The discriminator shown is a Convolution-BatchNorm-ReLU network.

[0026] In one embodiment, obtaining a target SAR principal axis prediction data set according to the predicted target parameters to be measured includes:

[0027] Acquire imaging angle data;

[0028] Inputting the predicted target parameters to be measured and the imaging angle data into SAR image simulation software for simulation to obtain a multi-angle simulated SAR image;

[0029] The multi-angle simulated SAR images constitute a target SAR principal axis prediction data set.

[0030] In one embodiment, it also includes:

[0031] The SAR image simulation software is RaySAR.

[0032] A SAR principal axis prediction dataset construction system based on a TransCGAN simulation parameter extraction framework is used to implement the SAR principal axis prediction dataset construction method based on a TransCGAN simulation parameter extraction framework as described above, comprising:

[0033] A simulation image acquisition module is used to acquire a target 3D model and obtain a simulated SAR image data set according to the target 3D model;

[0034] Prediction network building module, used to build a parameter prediction network based on the TransCGAN model;

[0035] A simulation parameter prediction module, used for acquiring a real SAR image, inputting the simulated SAR image data set and the real SAR image into the parameter prediction network, and obtaining predicted simulation parameters;

[0036] The target principal axis prediction module is used to obtain a target SAR principal axis prediction data set according to the predicted simulation parameters.

[0037] A device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the method for constructing a SAR principal axis prediction data set based on the TransCGAN simulation parameter extraction framework described in the above-mentioned embodiments are implemented.

[0038] A storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for constructing a SAR principal axis prediction data set based on the TransCGAN simulation parameter extraction framework described in each of the above embodiments.

[0039] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: the present invention can use the target 3D model and a single-angle real SAR image as the input data of the framework, obtain the mapping relationship between the SAR image and the simulation parameters by training the TransCGAN network, and use the RaySAR software to realize the conversion between the simulation parameters and the simulated SAR image, and finally realize the extraction of accurate simulation parameters from a single real SAR image and generate a large number of multi-angle simulated SAR image data sets for the training of the target main axis direction prediction network. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of a process of constructing a SAR principal axis prediction data set based on a TransCGAN simulation parameter extraction framework in one embodiment;

[0041] Figure 2 A schematic diagram of a target 3D model in an embodiment;

[0042] Figure 3 A schematic diagram of processing a 3D model in an embodiment;

[0043] Figure 4 A schematic diagram of a simulated SAR image data set in one embodiment;

[0044] Figure 5 A schematic diagram of a real SAR image in an embodiment;

[0045] Figure 6 A schematic diagram of a parameter prediction network structure in one embodiment;

[0046] Figure 7 A schematic diagram of a target SAR principal axis prediction data set in one embodiment;

[0047] Figure 8 A schematic diagram of the structure of a SAR principal axis prediction data set construction system based on a TransCGAN simulation parameter extraction framework in one embodiment;

[0048] Fig. 9 Schematic diagram of the internal structure of a device in one embodiment. DETAILED DESCRIPTION

[0049] Before describing the specific embodiments of the present invention, the overall concept of the present invention is described as follows:

[0050] The present invention is mainly developed for the process of predicting the main axis direction of synthetic aperture radar targets. With the rapid development of machine learning in the field of SAR image interpretation, the demand for high-quality simulated SAR image data sets for training and optimizing models is becoming increasingly urgent. In key tasks such as target detection, accurately predicting the main axis direction of targets in SAR images is of great significance, but it faces challenges such as data scarcity.

[0051] Therefore, the present invention proposes a SAR principal axis prediction dataset construction method based on the TransCGAN simulation parameter extraction framework, which can accurately extract simulation parameters from existing real SAR images and quickly generate a SAR target principal axis prediction dataset through simulation software.

[0052] After introducing the overall concept of the present invention, in order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below through specific implementation methods in conjunction with the accompanying drawings.

[0053] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of this specification should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of this specification do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0054] In one embodiment, Figure 1 As shown, a method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework is provided, including the following steps:

[0055] Step S101, obtaining a target 3D model, and obtaining a simulated SAR image data set according to the target 3D model.

[0056] Specifically, obtain the target 3D model, the target 3D model is as follows Figure 2 As shown, based on this target 3D model, a simulated SAR image dataset can be obtained.

[0057] On this basis, a target 3D model is obtained, and according to the target 3D model, a simulated SAR image data set is obtained, including:

[0058] Preprocessing the target 3D model to obtain a processed 3D model;

[0059] The processed 3D model is simulated by image simulation software to obtain a simulated SAR image data set.

[0060] Specifically, the target 3D model is preprocessed using 3D Studio Max software. To ensure the accuracy of subsequent simulation and the convenience of operation, the center of mass of the target object needs to be located at the center of the 3ds Max workspace. According to the actual application scenario, the size of the target model is adjusted and its direction setting in space is optimized.

[0061] Next, use the 3D model conversion tool at3D to further process the target 3D model. Considering that the simulation process needs to control the performance of the target surface characteristics through precisely set simulation parameters, it is necessary to erase the texture information in all layers of the target 3D model in advance to reduce interference factors and ensure that the electromagnetic scattering characteristics of the target 3D model can be accurately reflected based on the geometric properties of the target 3D model itself during the simulation process. After that, the processed 3D model is converted into a .pov format file to obtain a processed 3D model. This format is suitable for a variety of ray tracing renderers to ensure that the processed 3D model can obtain accurate simulation results after being input into the RaySAR software. Processing 3D models such as Figure 3 shown.

[0062] 3D Studio Max (3ds Max for short) and AccuTrans3D (at3d for short). 3ds Max is a PC-based software developed by Discreet that provides comprehensive 3D modeling, rendering and animation solutions. AccuTrans3D is a powerful 3D file conversion tool that supports more than 30 file types, provides accurate translation of 3D geometry, and can maintain 3D mesh information.

[0063] On this basis, the processed 3D model is simulated by image simulation software to obtain a simulated SAR image data set including:

[0064] Obtain the value range of specific simulation parameters and target parameters to be measured;

[0065] Inputting the specific simulation parameters and the processed 3D model into the image simulation software, sampling and taking values ​​of the target parameters to be measured one by one within the value range, so as to perform simulation and obtain a plurality of simulated SAR images;

[0066] The multiple simulated SAR images constitute the simulated SAR image dataset.

[0067] Specifically, the parameter prediction network based on the TransCGAN model in this embodiment mainly predicts several material surface parameters that are difficult to determine during image simulation but have a greater impact on the image, including the diffuse reflectance and roughness of the ground and the target, respectively. In addition to the target parameters to be measured, SAR image simulation using RaySAR also requires some other parameters, including the type and angle of light, the position of the camera, etc. These parameters need to be debugged and set in advance based on experience.

[0068] The interaction between each parameter variable and its impact on the simulation results are deeply analyzed. Based on this analysis, each target parameter to be tested is sampled one by one within a given range. For those target parameters to be tested that will cause significant visual changes to the generated image or have a significant impact on the output results, a dense sampling strategy is adopted to capture various potential changes caused by small changes in these target parameters to be tested, thereby enhancing the diversity and representativeness of the data set. For those target parameters to be tested whose numerical changes have little impact on the generated image, a sparse sampling strategy is adopted to reduce the amount of unnecessary data, which can not only ensure the validity of the data set, but also reasonably control resource consumption and avoid invalid accumulation of data.

[0069] When determining the value range of the sampling target parameter to be tested, it is necessary to ensure that its coverage is wide and reasonable, covering all possible expected situations as much as possible to prevent prediction failure problems during the training and testing stages. At the same time, the setting of the sampling interval is also particularly critical, especially in parameter-sensitive areas. The sampling interval is the smallest near the value most likely to have important changes, so as to better capture subtle change trends, thereby improving the generalization ability and accuracy of the prediction network.

[0070] Get the value ranges of specific simulation parameters (other simulation parameters except the parameters to be tested) and target parameters to be tested (target parameters to be tested generated by the parameter prediction network based on the TransCGAN model) obtained through debugging.

[0071] Input specific simulation parameters and processed 3D models into the image simulation software, sample the target parameters one by one within the value range to perform simulation, such as Figure 4 As shown, a series of corresponding simulated SAR images are generated, thereby constructing a simulated SAR image data set for predicting parameters.

[0072] Step S102, constructing a parameter prediction network based on the TransCGAN model.

[0073] Specifically, the parameter prediction network is based on the TransCGAN network. The main structure uses a conditional generative adversarial network (CGAN) with a generator and a discriminator. The generator uses the Swin Transformer structure, and the discriminator uses a hierarchical Convolution-BatchNorm-ReLU (CBR) structure. The Swin Transformer uses a self-attention mechanism to better handle long-distance dependencies and the construction of a shifted window, which enables the model to have a strong ability to capture global image information.

[0074] On this basis, the parameter prediction network based on the TransCGAN model is constructed including:

[0075] The parameter prediction network is shown below:

[0076]

[0077] Where L represents the loss function, G represents the generator, D represents the discriminator, min represents the minimum value, max represents the maximum value, V(D,G) represents the parameter prediction network, E[·] represents the expected function, x represents the real data, i.e., the real SAR image; y represents the conditional information, i.e., the simulated SAR image, and p z (z) represents the distribution of input noise, z represents the input noise of the generator, G(z,y) represents the predicted simulation parameters generated by the generator; D(x,y), D(G(z,y),y) represent the discrimination results of the discriminator, p data (x) represents the true distribution of samples in the data set, and log represents the logarithmic operation.

[0078] Step S103, obtaining a real SAR image, inputting the simulated SAR image data set and the real SAR image into the parameter prediction network, and obtaining predicted target parameters to be measured.

[0079] Specifically, the simulated SAR images are used as conditional inputs during training, and the real SAR images are used as conditions during testing. After the parameter prediction network is built, the simulated SAR image dataset is used to train the model. GAN optimizes the model through adversarial training between the generator and the discriminator, and the condition y is used as additional information input, which is extended to CGAN.

[0080] During training, the loss function of its generator G can be expressed as:

[0081]

[0082] The loss function of the discriminator D is expressed as:

[0083]

[0084] In the above loss function, x represents the real data, that is, the actual simulation parameters; y represents the conditional information, that is, the simulated SAR image; z represents the input noise of the generator, p z (z) represents the distribution of input noise; G represents the generator, G(z,y) represents the predicted simulation parameters generated by the generator; D(x,y) and D(G(z,y),y) represent the discriminant results; E[·] represents the expectation. During the training process, the discriminator and the generator are updated alternately to improve the performance by minimizing their respective loss functions. The overall objective function of the entire network is:

[0085]

[0086] The environmental conditions of the simulated SAR images used for training are consistent with those of the real SAR images used for testing. The real SAR images are obtained from the MSTAR dataset, which focuses on high-resolution imaging of a single target. There are multiple real SAR images in the dataset. A real SAR image is input into the trained parameter prediction network to obtain a set of predicted simulation parameters. Figure 5 As shown, the parameter prediction network structure is as follows Figure 6 shown.

[0087] Step S104, obtaining a target SAR principal axis prediction data set according to the predicted target parameters to be measured.

[0088] Specifically, the predicted target parameters to be measured and the specific simulation parameters are input into the SAR simulation software together to obtain a target SAR main axis prediction data set, which can be used to train the main axis direction prediction network.

[0089] On this basis, according to the predicted target parameters to be measured, the target SAR main axis prediction data set is obtained, including:

[0090] Acquire imaging angle data;

[0091] Inputting the predicted target parameters to be measured and the imaging angle data into SAR image simulation software for simulation to obtain a multi-angle simulated SAR image;

[0092] The multi-angle simulated SAR images constitute a target SAR principal axis prediction data set.

[0093] Specifically, multiple imaging angles are obtained, and the predicted target parameters to be measured, specific simulation parameters, and imaging angle data are input into the SAR image simulation software for simulation. The SAR image simulation software uses RaySAR software. The POV-Ray part of the RaySAR software is used to process the .pov file generated in step 1, and the corresponding predicted target parameters to be measured and specific simulation parameters are set. The optical 3D image and an output file "Contributions.txt" containing SAR-related information are rendered. Then the txt file is input into the MATLAB GUI of RaySAR, and different imaging angles are set for simulation, with the azimuth angle as the label, such as Figure 7 As shown, by obtaining multi-angle simulated SAR images, a target SAR main axis prediction data set can be constructed, which can be used to predict the SAR target main axis direction.

[0094] The RaySAR software used in the SAR image simulation software is an open source three-dimensional SAR simulation software designed by Auer of the Institute of Remote Sensing Technology of the German Aerospace Center. It is based on the physical optics model and ray tracing method. It is divided into the POV-Ray software part and the MATLAB GUI part. The main simulation parameters are set in the .pov file.

[0095] The present invention provides a method for constructing a SAR principal axis prediction data set based on the TransCGAN simulation parameter extraction framework. The method takes a target 3D model and a real SAR image at a single angle as input data of the framework, obtains the mapping relationship between the SAR image and the simulation parameters by training the TransCGAN network, and uses the RaySAR software to realize the conversion between the simulation parameters and the simulated SAR image. Finally, accurate simulation parameters are extracted from a single real SAR image and a large number of multi-angle simulated SAR image data sets are generated thereby, which are used for training the target principal axis direction prediction network.

[0096] Compared with the traditional SAR image simulation method, the present invention combines machine learning with a simulation method based on imaging principles and uses machine learning to assist SAR image simulation, which has faster simulation speed and better simulation effect, improves the utilization rate of real SAR image data, and enhances the interpretability of the simulation.

[0097] Furthermore, the parameter extraction network used in this solution is based on the TransCGAN model, which combines Swin Transformer with CGAN. CGAN can generate conditional sequence data by introducing real SAR images as conditional information and learning the relationship between conditional information and data. Swin Transformer uses a hierarchical feature expression method, and through the construction of self-attention and sliding windows, the model can more effectively process features of different scales and establish connections between different windows, thereby enhancing the model's ability to capture global information of the image. Combining the adversarial training method of GAN with the self-attention mechanism of Transformer can better model long-distance dependencies, maintain the intrinsic connection of data during the generation process, improve the training efficiency of the model and the quality of generated data, and enhance the stability of the model.

[0098] It should be noted that the method of the embodiment of the present invention can be performed by a single device, such as a computer or a server. The method of this embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present invention, and the multiple devices will interact with each other to complete the described method.

[0099] It should be noted that some embodiments of the present invention are described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0100] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present invention also provides a SAR principal axis prediction data set construction system based on the TransCGAN simulation parameter extraction framework.

[0101] refer to Figure 8 The SAR principal axis prediction data set construction system based on the TransCGAN simulation parameter extraction framework includes:

[0102] The simulated image acquisition module 801 is used to acquire a target 3D model and obtain a simulated SAR image data set according to the target 3D model;

[0103] A prediction network construction module 802 is used to construct a parameter prediction network based on the TransCGAN model;

[0104] A simulation parameter prediction module 803 is used to obtain a real SAR image, input the simulated SAR image data set and the real SAR image into the parameter prediction network, and obtain predicted simulation parameters;

[0105] The target principal axis prediction module 804 is used to obtain a target SAR principal axis prediction data set according to the predicted simulation parameters.

[0106] For the convenience of description, the above device is described as being divided into various modules according to their functions. Of course, when implementing the present invention, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0107] The device of the above embodiment is used to implement the corresponding SAR principal axis prediction data set construction method based on the TransCGAN simulation parameter extraction framework in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0108] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for constructing a SAR principal axis prediction data set based on the TransCGAN simulation parameter extraction framework as described in any of the above embodiments is implemented.

[0109] Fig. 9 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0110] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0111] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0112] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0113] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0114] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040).

[0115] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0116] The electronic device of the above embodiment is used to implement the corresponding SAR principal axis prediction data set construction method based on the TransCGAN simulation parameter extraction framework in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0117] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present invention also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the SAR principal axis prediction data set construction method based on the TransCGAN simulation parameter extraction framework as described in any of the above embodiments.

[0118] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0119] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the SAR principal axis prediction data set construction method based on the TransCGAN simulation parameter extraction framework as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0120] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0121] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0122] In addition, to simplify the description and discussion, and in order not to obscure the embodiments of the present invention, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, the devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present invention, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present invention will be implemented (i.e., these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present invention, it will be apparent to those skilled in the art that embodiments of the present invention may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0123] Although the invention has been described in conjunction with specific embodiments of the invention, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0124] The embodiments of the present invention are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework, characterized in that: include: Acquire a target 3D model, and obtain a simulated SAR image data set according to the target 3D model; Construct a parameter prediction network based on the TransCGAN model; Acquire a real SAR image, input the simulated SAR image data set and the real SAR image into the parameter prediction network, and obtain predicted target parameters to be measured; According to the predicted target parameters to be measured, a target SAR principal axis prediction data set is obtained.

2. The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework according to claim 1, characterized in that: The acquiring of the target 3D model and obtaining the simulated SAR image data set according to the target 3D model comprises: Preprocessing the target 3D model to obtain a processed 3D model; The processed 3D model is simulated by image simulation software to obtain a simulated SAR image data set.

3. The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework according to claim 2, characterized in that: The processing 3D model is simulated by image simulation software to obtain a simulated SAR image data set, which includes: Obtain the value range of specific simulation parameters and target parameters to be measured; Inputting the specific simulation parameters and the processed 3D model into the image simulation software, sampling and taking values ​​of the target parameters to be measured one by one within the value range, so as to perform simulation and obtain a plurality of simulated SAR images; The multiple simulated SAR images constitute the simulated SAR image dataset.

4. The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework according to claim 1, characterized in that: The construction of a parameter prediction network based on the TransCGAN model includes: The parameter prediction network is as follows: Where L represents the loss function, G represents the generator, D represents the discriminator, min represents the minimum value, max represents the maximum value, V(D,G) represents the parameter prediction network, E[·] represents the expected function, x represents the real data, i.e., the real SAR image; y represents the conditional information, i.e., the simulated SAR image, and p z (z) represents the distribution of input noise, z represents the input noise of the generator, G(z,y) represents the predicted simulation parameters generated by the generator; D(x,y), D(G(z,y),y) represent the discrimination results of the discriminator, p data (x) represents the true distribution of samples in the data set, and log represents the logarithmic operation.

5. The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework according to claim 4, characterized in that: Also includes: The generator is a Swin Transformer network based on the attention mechanism; The discriminator shown is a Convolution-BatchNorm-ReLU network.

6. The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework according to claim 1, characterized in that: The step of obtaining a target SAR principal axis prediction data set according to the predicted target parameters to be measured comprises: Acquire imaging angle data; Inputting the predicted target parameters to be measured and the imaging angle data into SAR image simulation software for simulation to obtain a multi-angle simulated SAR image; The multi-angle simulated SAR images constitute a target SAR principal axis prediction data set.

7. The method for constructing a SAR principal axis prediction dataset based on the TransCGAN simulation parameter extraction framework according to claim 6, characterized in that: Also includes: The SAR image simulation software is RaySAR.

8. A SAR principal axis prediction dataset construction system based on the TransCGAN simulation parameter extraction framework, characterized in that: The method for constructing a SAR principal axis prediction data set based on the TransCGAN simulation parameter extraction framework as described in claims 1 to 7 comprises: A simulation image acquisition module is used to acquire a target 3D model and obtain a simulated SAR image data set according to the target 3D model; Prediction network building module, used to build a parameter prediction network based on the TransCGAN model; A simulation parameter prediction module, used for acquiring a real SAR image, inputting the simulated SAR image data set and the real SAR image into the parameter prediction network, and obtaining predicted simulation parameters; The target principal axis prediction module is used to obtain a target SAR principal axis prediction data set according to the predicted simulation parameters.

9. A device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images