Coding method and device for radar image-oriented sensing fusion system

Abstract

The invention discloses a coding method and device for a radar image-oriented sensing fusion system. The method comprises the following steps of: S1, acquiring a data set; s2, establishing a semantic encoder model, wherein the semantic encoder model comprises a backbone network module and a neck network module; s3, establishing a neural network channel encoder model, wherein the neural network channel encoder model comprises a multi-scale fusion module and a redundancy increasing module; and S4, loading the feature information output by the neural network channel encoder model to a physical channel and sending the feature information. According to the scheme, the integrated design of the radar image communication and perception technology based on target detection is realized, and the semantic encoder model can effectively realize the fusion of the target detection technology and the semantic communication technology; and through the improved algorithm, the data has better robustness to noise interference.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and in particular relates to a coding method and a device for a radar image synaesthesia fusion system. Background Art

[0002] Synthetic aperture radar (SAR) has a unique imaging capability: it can provide high-resolution two-dimensional images without being affected by sunlight, cloud coverage and weather conditions. In recent years, SAR has important research significance in many fields such as marine environmental monitoring, resource exploration and disaster emergency response, and object detection in SAR images has a wide range of application scenarios. The combination of SAR image target recognition and communication system can achieve efficient and reliable detection and early warning, and realize information synaesthesia integration. The current research on SAR image recognition system and communication system is separated, and the combination of the two is not effective enough. The emergence of semantic communication technology provides an effective design idea for the integration of SAR system and communication system.

[0003] At present, radar image communication and perception technologies based on target detection are designed separately. They divide the communication into two parts, either detection first and then transmission, or transmission first and then detection. They do not consider designing an integrated network architecture.

[0004] In the prior art, an invention patent with a publication number of CN117036975A discloses a method and device for target recognition of SAR images, the method comprising: obtaining a SAR image to be recognized; extracting quantitative semantic features and image features of the SAR image to be recognized; mapping the quantitative semantic features of the SAR image to be recognized into qualitative semantic features of the SAR image to be recognized based on a pre-constructed semantic mapping association between quantitative semantic features and qualitative semantic features; respectively encoding the image features, quantitative semantic features and qualitative semantic features of the SAR image to be recognized, and performing feature fusion on the respectively encoded features, so as to output a target recognition result of the SAR image to be recognized according to the feature fusion result.

[0005] Another invention patent with publication number CN118587439A discloses a multi-source remote sensing image semantic segmentation method and device based on a denoising diffusion probability model, including a Mamba-based dual-branch backbone network and a denoising diffusion probability model network; the Mamba-based dual-branch backbone network extracts features and fuses features of paired original optical images and original synthetic aperture radar images in four stages to obtain multi-source fusion features of four stages, and then splices the multi-source fusion features of the four stages into a multi-scale fusion feature; in the denoising diffusion probability model network, the original semantic segmentation label is denoised by a forward denoising module, and the result of splicing the multi-scale fusion feature and the noisy original semantic segmentation label is used as the output of the forward denoising module; the input feature is denoised by a noise decoder to predict the semantic segmentation label of the input feature.

[0006] Another invention patent with publication number CN118540024A discloses a semantic communication encoding and decoding method, device, equipment and storage medium, the method comprising: extracting semantic features of target data according to a semantic encoder; converting the semantic features into a preset data transmission format according to a channel encoder; transmitting the semantic features to a receiving end according to a channel transmission module; decoding the semantic features according to a channel decoder to obtain the target data; converting the target data back to the original data format according to the semantic decoder; wherein the semantic encoder and the semantic decoder are trained according to a deep learning model, and the semantic encoder, channel encoder, channel transmission module, channel decoder and semantic encoder are taken as a whole, and a preset convolutional neural network deep learning model is trained according to the overall and end-to-end training method. Summary of the invention

[0007] A brief summary of embodiments of the present invention is provided below to provide a basic understanding of certain aspects of the present invention. It should be understood that the following summary is not an exhaustive summary of the present invention. It is not intended to identify key or important parts of the present invention, nor is it intended to limit the scope of the present invention. Its purpose is merely to provide certain concepts in a simplified form as a prelude to a more detailed description discussed later.

[0008] To solve the above problems, the present application provides a coding method and device for a radar image synaesthesia fusion system, which can form an integrated structure of transmission and detection.

[0009] According to one aspect of the present application, a coding method for a radar image synaesthesia fusion system is provided, comprising the following steps: Step S1: Acquire a data set, which includes a synthetic aperture radar image data set, a test environment parameter, and a signal-to-noise ratio parameter; Step S2: establishing a semantic encoder model, which includes a backbone network module and a neck network module, and is used to extract multi-layer deep feature information of a synthetic aperture radar image dataset; Step S3: establishing a neural network channel encoder model, the neural network channel encoder model including a multi-scale fusion module and a redundancy increase module, for processing the multi-layer deep feature information output by the semantic encoder model and outputting the feature information; Step S4: Load the feature information output by the neural network channel encoder model onto the physical channel and send it.

[0010] As a specific solution, in step S1, the synthetic aperture radar image data set includes basic parameters such as size and input batch. In addition, the signal-to-noise ratio parameter can select the target signal-to-noise ratio of the design process according to the actual scenario.

[0011] As a specific solution, in step S2, the semantic encoder model is established based on the backbone network (Backbone) module and the neck network (Neck) module of the single target detection framework; The backbone network module includes a first convolutional layer CBS, a second convolutional layer CBS, a first cross-stage partial bottleneck convolution C2f, a third convolutional layer CBS, a second cross-stage partial bottleneck convolution C2f, a fourth convolutional layer CBS, a third cross-stage partial bottleneck convolution C2f, a fifth convolutional layer CBS, a fourth cross-stage partial bottleneck convolution C2f and a spatial pyramid fast pooling SPPF, which are connected in sequence; The neck network (Neck) module includes a first upsampling operation Upsample, a first scale splicing operation Concat, a fifth cross-stage partial bottleneck convolution C2f, a second upsampling operation Upsample, a second scale splicing operation Concat, a sixth cross-stage partial bottleneck convolution C2f, a sixth convolutional layer CBS, a third scale splicing operation Concat, a seventh cross-stage partial bottleneck convolution C2f, a seventh convolutional layer CBS, a fourth Concat and an eighth cross-stage partial bottleneck convolution C2f, which are connected in sequence.

[0012] In the above scheme, the backbone module adopts a series of convolution and deconvolution layers (CBS), and uses residual connections and bottleneck structures to reduce the size of the network and improve performance. The backbone module adopts a cross-stage partial bottleneck convolution (C2f) module based on the idea of ​​cross-stage local network, which makes the single target detection framework more lightweight and can obtain richer gradient flow information. The spatial pyramid fast pooling module is used at the end of the backbone module. The spatial pyramid fast pooling (SPPF) module passes three Maxpools of size 5×5 in series, and then splices each layer to perform multi-scale feature fusion, so that the model can use global and local information at the same time, thereby improving the ability to locate and identify targets while greatly reducing the computational overhead of the network.

[0013] This application introduces spatial pyramid fast pooling SPPF and cross-stage partial bottleneck convolution C2f for radar image application scenarios, making the Backbone module more lightweight and obtaining richer feature information. Compared with the existing SPP, SPPF is more lightweight and improves the model's ability to obtain local and global feature information, greatly improving the ability to locate and identify targets.

[0014] The neck network (Neck) module uses the path aggregation network feature pyramid network (PAN-FPN) structure. PAN-FPN constructs a top-down and bottom-up network structure, and realizes the complementarity of shallow position information and deep semantic information through feature fusion, thereby realizing the diversity and completeness of feature information, and further enhancing the representation ability of features. For the current application scenario, the Neck module adopts the path aggregation network feature pyramid structure (PAN-FPN). PAN-FPN is a network structure that combines top-down and bottom-up. Compared with the existing bottom-up (PAN) and top-down (FPN), PAN-FPN can realize the complementarity of front position information and deep semantic information, thereby realizing the diversity and completeness of feature information, and further enhancing the expression ability of features.

[0015] Among them, each convolutional layer CBS (the first convolutional layer CBS...the seventh convolutional layer CBS) has the same structure, including sequentially linked convolution, normalization processing and activation function.

[0016] Spatial pyramid fast pooling SPPF includes CBS, the first maximum pooling, the second maximum pooling, the third maximum pooling, scale splicing and CBS connected in sequence.

[0017] Each cross-stage partial bottleneck convolution C2f (the first cross-stage partial bottleneck convolution C2f ... the seventh cross-stage partial bottleneck convolution C2f) has the same structure, including CBS, scale separation, the first bottleneck layer, the second bottleneck layer, the third bottleneck layer, scale splicing and CBS connected in sequence.

[0018] The upsample upsampling operation and the Concat scale feature concatenation operation both use existing technologies, so they are not described in detail here.

[0019] As a specific solution, in step S2, the multi-layer depth feature information of the synthetic aperture radar image data set is extracted by the semantic encoder model and expressed by the following formula: Assume that the synthetic aperture radar image dataset input is an RGB image , where M is the input image, R is the RGB image in the synthetic aperture radar image dataset, H represents the height of the RGB image, W represents the width of the RGB image, and 3 represents the number of channels of the RGB image is 3; the result of the input data after passing through the semantic encoding model is S, which is expressed as: (1) Among them, S is the output of the semantic encoder, Represents the semantic coding network, whose parameter is β, β includes the first convolution layer CBS, the second convolution layer CBS, the first cross-stage partial bottleneck convolution C2f, the third convolution layer CBS, the second cross-stage partial bottleneck convolution C2f, the fourth convolution layer CBS, the third cross-stage partial bottleneck convolution C2f, the fifth convolution layer CBS, the fourth cross-stage partial bottleneck convolution C2f and spatial pyramid fast pooling SPPF of the backbone network module, and the first upsampling operation Upsample, the first scale splicing operation Concat, the fifth cross-stage partial bottleneck convolution C2f, the second upsampling operation Upsample, the second scale splicing operation Concat, the sixth cross-stage partial bottleneck convolution C2f, the sixth convolution layer CBS, the third scale splicing operation Concat, the seventh cross-stage partial bottleneck convolution C2f, the seventh convolution layer CBS, the fourth Concat and the eighth cross-stage partial bottleneck convolution C2f of the neck network module. The synthetic aperture radar image dataset includes the STCD dataset, image dataset, etc.

[0020] As a specific solution, step S3 specifically includes: Step S31: Establish a multi-scale fusion module to fuse the multi-scale feature information (M1, M2, M3) transmitted by the neck network (Neck) module to generate new feature information X1; the specific fusion process is as follows: (2) in is the upsampling operation; represents the concatenation operation; M1, M2, and M3 are the outputs of the three layers of the semantic encoding module Neck network at different scales, namely, the outputs of the sixth cross-stage partial bottleneck convolution C2f, the seventh cross-stage partial bottleneck convolution C2f, and the eighth cross-stage partial bottleneck convolution C2f; Step S32: The new characteristic information X1 generated in step S31 is first passed through a redundancy adding module and then normalized to obtain normalized characteristic information X, so as to ensure that the signal maintains its quality and integrity as much as possible during the transmission process; the specific process is as follows: , (3) In the formula, X is the normalized feature information, (•) indicates the redundancy layer operation of the redundancy increase module, and its parameters are ; BN (•) is the normalization operation.

[0021] The multi-scale fusion module is used to effectively realize the fusion and information transmission of the semantic encoder model and the neural network channel encoder model.

[0022] As a specific solution, step S4 specifically includes: Step S41: establishing a physical channel, where the physical channel adopts a Gaussian white noise channel or a Rayleigh fading channel; Step S42: The normalized characteristic information X is transmitted through the physical channel established in step S41, and the signal Y received at the receiving end is expressed as: (4) Where Y is the signal received at the receiving end; h is the Rayleigh fading channel gain, η is the mean value of 0 and the variance is σ 2 Gaussian noise, that is, η~(0,σ 2 ) (Gaussian noise η follows a normal distribution with a mean of 0 and a variance of σ2); for AWGN channel is the channel model representation under Gaussian channel, for fading channel is the channel model representation under Rayleigh fading channel.

[0023] According to another aspect of the present application, a coding device for a synaesthesia fusion system for radar images is provided, which is implemented by using the above coding method.

[0024] This application adopts the above solution, which has the following advantages compared with the prior art: 1. Currently, radar image communication and perception technologies based on target detection are designed separately, and there is no integrated design yet; 2. The semantic encoder of this encoding method adopts the combination of Backbone network and Neck network modules, which can effectively realize the integration of target detection technology and semantic communication technology; 3. The neural network-based channel encoder designed by the coding method consists of a multi-scale fusion module and a redundancy increase module. The designed multi-scale fusion module can realize the organic coupling of the semantic encoder and the channel encoder.

[0025] In addition, through the above improved algorithm, experiments have proved that as the signal-to-noise ratio increases, the data becomes more and more stable, and the data has better robustness to noise interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The present invention may be better understood by referring to the following description given in conjunction with the accompanying drawings, wherein the same or similar reference numerals are used throughout the drawings to represent the same or similar components. The accompanying drawings, together with the following detailed description, are included in and form a part of this specification and are used to further illustrate the preferred embodiments of the present invention and to explain the principles and advantages of the present invention. In the drawings: Figure 1 This is a structural diagram of an encoding method for a radar image synaesthesia fusion system according to an embodiment of the present invention; Figure 2 A diagram of the structure of a semantic encoder according to an embodiment of the present invention; Figure 3 Backbone and Neck structure diagram of an embodiment of the present invention; Figure 4 A structural diagram of a CBS module according to an embodiment of the present invention; Figure 5 The SPPF module structure diagram of the embodiment of the present invention; Figure 6 C2f structure diagram of an embodiment of the present invention; Figure 7 The PAN-FPN structure diagram of the embodiment of the present invention, the direction indicated by the arrow is the direction of data flow; Figure 8 A structural diagram of a channel encoder based on a neural network according to an embodiment of the present invention; Fig. 9 A redundancy increasing layer structure diagram of an embodiment of the present invention; Fig.10 Ten groups of data at different SNRs according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] Embodiments of the present invention will be described below with reference to the accompanying drawings. Elements and features described in one drawing or one embodiment of the present invention may be combined with elements and features shown in one or more other drawings or embodiments. It should be noted that for the purpose of clarity, the representation and description of components and processes that are not related to the present invention and are known to those of ordinary skill in the art are omitted in the drawings and descriptions.

[0028] The present invention provides a coding method for a radar image synaesthesia fusion system, which is characterized by comprising the following steps: Step S1: initialization phase, including determination of input synthetic aperture radar image data set, test environment parameters, signal-to-noise ratio parameters and important parameters in the corresponding design process; Step S2: Establishing a semantic encoder model, which mainly includes a backbone network module and a neck network module, for extracting multi-layer deep feature information of radar image data; Step S3: According to the multi-layer deep feature information obtained by the semantic coding model, it is passed to a channel encoder model based on a neural network, which mainly includes a multi-scale fusion module and a redundancy increase module; Step S4: Load the feature information output by the neural network-based channel encoder onto the physical channel and send it.

[0029] Wherein, step S1 is specifically as follows: Step S11: Determine the basic parameters of the synthetic aperture radar image data set including size, input batch and important parameters in the corresponding design process.

[0030] Step S12: Determine the target signal-to-noise ratio of the design process according to the scenario.

[0031] Step S2 is specifically as follows: Based on the backbone network (Backbone) module and the neck network (Neck) module of the single target detection framework, a basic semantic encoder model is established. The framework diagram is shown in Figure 2 .

[0032] The detailed model diagrams of the backbone network module and the neck network module are shown in Figure 3 The backbone network module includes the first convolution layer CBS, the second convolution layer CBS, the first cross-stage partial bottleneck convolution C2f, the third convolution layer CBS, the second cross-stage partial bottleneck convolution C2f, the fourth convolution layer CBS, the third cross-stage partial bottleneck convolution C2f, the fifth convolution layer CBS, the fourth cross-stage partial bottleneck convolution C2f and the spatial pyramid fast pooling SPPF, which are connected in sequence.

[0033] The neck network (Neck) module includes a first upsampling operation Upsample, a first scale splicing operation Concat, a fifth cross-stage partial bottleneck convolution C2f, a second upsampling operation Upsample, a second scale splicing operation Concat, a sixth cross-stage partial bottleneck convolution C2f, a sixth convolution layer CBS, a third scale splicing operation Concat, a seventh cross-stage partial bottleneck convolution C2f, a seventh convolution layer CBS, a fourth Concat and an eighth cross-stage partial bottleneck convolution C2f, which are connected in sequence; Among them, each convolutional layer CBS (the first convolutional layer CBS...the seventh convolutional layer CBS) has the same structure, including sequentially linked convolution, normalization processing and activation function.

[0034] Spatial pyramid fast pooling SPPF includes CBS, the first maximum pooling, the second maximum pooling, the third maximum pooling, scale splicing and CBS connected in sequence.

[0035] Each cross-stage partial bottleneck convolution C2f (the first cross-stage partial bottleneck convolution C2f ... the seventh cross-stage partial bottleneck convolution C2f) has the same structure, including CBS, scale separation, the first bottleneck layer, the second bottleneck layer, the third bottleneck layer, scale splicing and CBS connected in sequence.

[0036] The upsample upsampling operation and the Concat scale feature concatenation operation both use existing technologies, so they are not described in detail here.

[0037] The backbone module is responsible for feature extraction. It downsamples the input radar image five times using a series of convolution and deconvolution layers to obtain five different scale features. It also uses residual connections and bottleneck structures to reduce the size of the network and improve performance. The backbone module uses a cross-stage partial bottleneck convolution (C2f) module based on the idea of ​​a cross-stage local network, which makes the single target detection framework more lightweight while also obtaining richer gradient flow information. The spatial pyramid fast pooling module (SPPF) is used at the end of the backbone module. The spatial pyramid fast pooling module passes three 5×5 Maxpools in series, and then splices each layer to perform multi-scale feature fusion, so that the model can use global and local information at the same time, thereby improving the ability to locate and identify targets while greatly reducing the computational overhead of the network.

[0038] Step S23: The path aggregation network feature pyramid network (PAN-FPN) structure is used in the design of the neck network (Neck) module. PAN-FPN constructs a top-down and bottom-up network structure, and realizes the complementarity of shallow position information and deep semantic information through feature fusion, thereby realizing the diversity and completeness of feature information and further enhancing the representation ability of features.

[0039] The processing of step S2 can be expressed by formula 1. Assume that the image input of the STCD dataset is an RGB image , the result after the input data passes through the semantic encoding model is .

[0040] (1) Among them, S is the output of the semantic encoder. Figure 3 The positions in are the outputs of the sixth cross-stage partial bottleneck convolution C2f, the seventh cross-stage partial bottleneck convolution C2f, and the eighth cross-stage partial bottleneck convolution C2f; Represents the semantic encoding network, whose parameter is β, β represents the convolution layer, BN normalization, and activation function. Specifically, β is the backbone network module (the first convolution layer CBS, the second convolution layer CBS, the first cross-stage partial bottleneck convolution C2f, the third convolution layer CBS, the second cross-stage partial bottleneck convolution C2f, the fourth convolution layer CBS, the third cross-stage partial bottleneck convolution C2f, the fifth convolution layer CBS, the fourth cross-stage partial bottleneck convolution C2f and spatial pyramid fast pooling SPPF) and the neck network module (the first upsampling operation Upsample, The first scale splicing operation Concat, the fifth cross-stage partial bottleneck convolution C2f, the second upsampling operation Upsample, the second scale splicing operation Concat, the sixth cross-stage partial bottleneck convolution C2f, the sixth convolution layer CBS, the third scale splicing operation Concat, the seventh cross-stage partial bottleneck convolution C2f, the seventh convolution layer CBS, the fourth Concat and the eighth cross-stage partial bottleneck convolution C2f) are superimposed on the convolution layer CBS, the cross-stage partial bottleneck convolution C2f, the spatial pyramid fast pooling SPPF and other modules. The bold S on the left side of the equation is the output of the semantic encoder.

[0041] This application introduces the spatial pyramid fast pooling SPPF and cross-stage partial bottleneck convolution C2f for the application scenario of radar images, making the Backbone module more lightweight and obtaining richer feature information. Compared with the existing SPP, SPPF is more lightweight and improves the model's ability to obtain local and global feature information, greatly improving the ability to locate and identify targets. For the current application scenario, the Neck module adopts the path aggregation network feature pyramid structure (PAN-FPN). PAN-FPN is a network structure that combines top-down and bottom-up. Compared with the existing bottom-up (PAN) and top-down (FPN), PAN-FPN can achieve the complementarity of front position information and deep semantic information, thereby achieving the diversity and completeness of feature information, and further enhancing the expression ability of features.

[0042] Step S3 is specifically as follows: Step S31: A radar image multi-scale fusion module is established to combine the multi-scale feature information ( , , ) to generate new feature information By designing this module, the fusion and information transmission of the semantic coding module and the channel coding module can be effectively realized. The specific fusion process is as follows: (2) in is the upsampling operation; Represents a splicing operation; , , It is the output of the three layers of the semantic encoder Neck network at different scales, namely, the output of the sixth cross-stage partial bottleneck convolution C2f, the seventh cross-stage partial bottleneck convolution C2f, and the eighth cross-stage partial bottleneck convolution C2f.

[0043] Step S32: The feature information generated in step S31 Normalization is performed to ensure that the signal maintains its quality and integrity as much as possible during transmission. The specific process is as follows: , (3) In the formula, (•) indicates redundant layer operation, whose parameters are ; BN (•) is the normalization operation.

[0044] Specifically, the overall model diagram of step S3 is as follows: Figure 8 shown.

[0045] Step S4 is specifically as follows: Step S41: Establish a physical channel, which can be divided into a Gaussian white noise channel and a Rayleigh fading channel.

[0046] Step S42: The characteristic information X generated in step S32 is transmitted through the physical transmission channel established in step S41. Assuming that X is transmitted through the physical channel, the signal received at the receiving end can be given by Y, which can be expressed as (4) Where h is the Rayleigh fading channel gain, η is the mean value of 0, and the variance is σ 2 Gaussian noise, that is, η~(0,σ 2 ); for AWGN channel is the channel model representation under Gaussian channel, for fading channel is the channel model representation under Rayleigh fading channel.

[0047] The experimental environment of the embodiment of the present application is configured with Windows 10 operating system, the processor is Intel 13th generation Core i9-13900K, the graphics card is NVIDIA GeForceRTX4090, 24GB RAM, Pytorch version is 2.1.1, and CUDA version is 12.1. The radar dataset used is the Sonar Common Target Detection Dataset (SCTD).

[0048] The specific data transmission process of this design invention can be simplified into the following steps, assuming that the STCD dataset image input is an RGB image , the result after the input data passes through the semantic coding model is S, and the result after the semantic coding output S is encoded by the channel encoder based on the neural network is X.

[0049] The result data were obtained when the signal-to-noise ratio was 0, 5, 10, 15, and 25 dB, and ten groups of X data were taken under each signal-to-noise ratio. The results are shown in Table 1.

[0050] Table 1 Data of ten groups of X under different SNR

[0051] It can be seen from the data in Table 1 that as the signal-to-noise ratio increases, the data becomes more and more stable. Fig.10 It can be seen intuitively that with the increase of signal-to-noise ratio, the data has better robustness to noise interference. This shows the rationality of the encoding method of the radar image synaesthesia fusion system we proposed, and further proves the effectiveness of our proposed algorithm.

[0052] At present, radar image communication and perception technologies based on target detection are designed separately, and there is no integrated design. This patent designs an encoding method for a radar image synaesthesia fusion system.

[0053] The semantic encoder of this encoding method adopts a combination of two modules, Backbone network and Neck network, which can effectively realize the integration of target detection technology and semantic communication technology.

[0054] The neural network-based channel encoder designed by the coding method consists of a multi-scale fusion module and a redundancy increase module. The designed multi-scale fusion module can realize the organic coupling of the semantic encoder and the channel encoder.

[0055] It should be emphasized that the term “include / comprises” when used herein refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0056] Although the present invention has been disclosed above by describing specific embodiments of the present invention, it should be understood that all the above embodiments and examples are exemplary rather than restrictive. Those skilled in the art may design various modifications, improvements or equivalents of the present invention within the spirit and scope of the appended claims. These modifications, improvements or equivalents should also be considered to be included in the protection scope of the present invention.

Claims

1. A coding method for a radar image synaesthesia fusion system, characterized by: include: Step S1: Acquire a data set, which includes a synthetic aperture radar image data set, a test environment parameter, and a signal-to-noise ratio parameter; Step S2: establishing a semantic encoder model, the semantic encoder model comprising a backbone network module and a neck network module, for extracting multi-layer deep feature information of a synthetic aperture radar image dataset; Step S3: establishing a neural network channel encoder model, the neural network channel encoder model including a multi-scale fusion module and a redundancy increase module, for processing the multi-layer deep feature information output by the semantic encoder model and outputting the feature information; Step S4: Load the feature information output by the neural network channel encoder model onto the physical channel and send it.

2. The encoding method of the synaesthesia fusion system according to claim 1, characterized in that: In the step S1, the synthetic aperture radar image dataset includes a size and an input batch.

3. The encoding method of the synaesthesia fusion system according to claim 1, characterized in that: In step S2, the semantic encoder model is established based on the backbone network module and the neck network module of the single target detection framework; The backbone network module includes a first convolutional layer CBS, a second convolutional layer CBS, a first cross-stage partial bottleneck convolution C2f, a third convolutional layer CBS, a second cross-stage partial bottleneck convolution C2f, a fourth convolutional layer CBS, a third cross-stage partial bottleneck convolution C2f, a fifth convolutional layer CBS, a fourth cross-stage partial bottleneck convolution C2f and a spatial pyramid fast pooling SPPF connected in sequence; The neck network module includes a first upsampling operation Upsample, a first scale splicing operation Concat, a fifth cross-stage partial bottleneck convolution C2f, a second upsampling operation Upsample, a second scale splicing operation Concat, a sixth cross-stage partial bottleneck convolution C2f, a sixth convolutional layer CBS, a third scale splicing operation Concat, a seventh cross-stage partial bottleneck convolution C2f, a seventh convolutional layer CBS, a fourth Concat and an eighth cross-stage partial bottleneck convolution C2f, which are connected in sequence.

4. The encoding method of the synaesthesia fusion system according to claim 3, characterized in that: In step S2, the multi-layer depth feature information of the synthetic aperture radar image data set is extracted by the semantic encoder model and expressed as follows: Assume that the input of the synthetic aperture radar image dataset is an RGB image , where M is the input image, R is the RGB image in the synthetic aperture radar image dataset, H represents the height of the RGB image, W represents the width of the RGB image, and 3 represents the number of channels of the RGB image is 3; the result of the input data after passing through the semantic encoding model is S, which is expressed as: ； Where S is the output of the semantic encoder, Represents the semantic encoding network, whose parameter is β, β includes the first convolution layer CBS, the second convolution layer CBS, the first cross-stage partial bottleneck convolution C2f, the third convolution layer CBS, the second cross-stage partial bottleneck convolution C2f, the fourth convolution layer CBS, the third cross-stage partial bottleneck convolution C2f, the fifth convolution layer CBS, the fourth cross-stage partial bottleneck convolution C2f and the spatial pyramid fast pooling SPPF of the backbone network module, and the first upsampling operation Upsample, the first scale splicing operation Concat, the fifth cross-stage partial bottleneck convolution C2f, the second upsampling operation Upsample, the second scale splicing operation Concat, the sixth cross-stage partial bottleneck convolution C2f, the sixth convolution layer CBS, the third scale splicing operation Concat, the seventh cross-stage partial bottleneck convolution C2f, the seventh convolution layer CBS, the fourth Concat and the eighth cross-stage partial bottleneck convolution C2f of the neck network module.

5. The encoding method of the synaesthesia fusion system according to claim 4, characterized in that: Step S3 specifically includes: Step S31: Establish a multi-scale fusion module to fuse the multi-scale feature information (M1, M2, M3) transmitted by the neck network module to generate new feature information X1; the specific fusion process is as follows: ； in is the upsampling operation; represents the concatenation operation; M1, M2, and M3 are the outputs of the sixth cross-stage partial bottleneck convolution C2f, the seventh cross-stage partial bottleneck convolution C2f, and the eighth cross-stage partial bottleneck convolution C2f of the semantic encoder, respectively; Step S32: The characteristic information X1 is first normalized by the redundancy adding module to obtain the normalized characteristic information X, so as to ensure that the signal maintains its quality and integrity as much as possible during the transmission process; the specific process is as follows: ； In the formula, X is the normalized feature information, (•) indicates redundant layer operation, whose parameters are ; BN (•) is the normalization operation.

6. The encoding method of the synaesthesia fusion system according to claim 1, characterized in that: The step S4 specifically includes: Step S41: establishing a physical channel, where the physical channel adopts a Gaussian white noise channel or a Rayleigh fading channel; Step S42: The normalized characteristic information X is transmitted through the physical channel established in step S41, and the signal Y received at the receiving end is expressed as: （4） Where Y is the signal received at the receiving end; h is the Rayleigh fading channel gain, η is the mean value of 0 and the variance is σ 2 Gaussian noise, that is, η~(0,σ 2 ) ; for AWGN channel is the channel model representation under Gaussian channel, for fading channel is the channel model representation under Rayleigh fading channel.

7. An encoding device for a radar image synaesthesia fusion system, characterized in that: The device executes the steps of the encoding method of the synaesthesia fusion system as described in any one of claims 1-6.

Images