A dual RIS-assisted radio complementary mode imaging system and method

Abstract

The present invention belongs to the field of wireless imaging technology and discloses a dual-RIS-assisted radio complementary mode imaging system and method, comprising a transmitter Tx, a first receiver Rx1, a second receiver Rx2, a forward RIS, a backward RIS, and a deep learning network. The forward RIS performs beam steering on the signal transmitted by Tx. A portion of the steered beam passing through the target sensing area (SoI) is reflected by the target, forming a reflected signal that is received by Rx1, while the remaining portion passes through the SoI and is received by the backward RIS as a transmitted signal, which is then beam-steeringly focused on Rx2. The deep learning network is used to extract features from the reflected and transmitted signals, correspondingly obtaining features for characterizing the reflection and transmission coefficients. After feature fusion, a contour image of the target is generated. The forward and backward RIS utilize pre-configured beamforming control methods when collecting electromagnetic data, sequentially scanning and sensing the space. This invention can improve the quality of reconstructed images.

Description

Technical Field

[0001] The present invention belongs to the field of wireless imaging technology, and more particularly, relates to an imaging system and method of a dual RIS-assisted radio complementary mode. Background Art

[0002] Radio frequency (RF) imaging is becoming increasingly popular in the field of electromagnetic (EM) sensing. RF imaging technology has demonstrated its ability to achieve high-resolution imaging under varying lighting conditions and in all weather conditions. It has applications in environmental monitoring, medical monitoring, autonomous driving, and security inspections. Traditional RF imaging systems rely on frequency diversity or spatial diversity for accurate sensing. Specifically, specific signal modulation techniques, such as ultra-wideband (UWB), linear frequency modulation (LFM), and orthogonal frequency division multiplexing (OFDM), broaden the bandwidth of the imaging signal and increase the degree of freedom of perception through frequency diversity. Furthermore, antenna diversity techniques, such as multiple-input, multiple-output (MIMO) antennas, can achieve super-resolution imaging in the angular domain, providing significant spatial diversity. While these operations of expanding signal bandwidth and increasing the number of RF antenna elements can improve image resolution, they also come with challenges such as increased power consumption, complexity, and cost of the RF transceiver system.

[0003] Reconfigurable smart surfaces (RIS), as an active member of the metamaterial family, have the advantages of low energy consumption, cost-effectiveness, and easy deployment. Combining RIS-based passive antennas with computational imaging (CI) expands the degrees of freedom of imaging.

[0004] Several studies have explored single-RIS-assisted reflection-mode imaging systems, demonstrating their feasibility by improving the multiplexing capability of spatial information. However, the complexity of the target structure and the influence of surface roughness relative to the illumination wavelength often lead to multiple scattering, generating weak and complex scattered waves. Extracting information from this complex, low-signal-to-noise reflection signal is challenging, especially from highly lossy objects. Such imaging data often results in blurred and incomplete image output from reconstruction algorithms. Summary of the Invention

[0005] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides an imaging system and method of a dual RIS-assisted radio complementary mode, the purpose of which is to improve imaging quality.

[0006] To achieve the above objectives, according to a first aspect of the present invention, a dual-RIS-assisted radio complementary mode imaging system is provided, comprising: a transmitter Tx, a first receiver Rx1, a second receiver Rx2, a forward RIS and a backward RIS placed oppositely on either side of a target sensing area SoI, and an imaging network based on deep learning;

[0007] The transmitter Tx is directed toward the forward RIS and is used to transmit a linear frequency modulation wave signal;

[0008] The forward RIS is configured to perform beam steering on the received linear frequency modulation wave signal so that the steered beam passes through the target sensing area SoI. A portion of the beam passing through the target sensing area SoI is reflected by the target to form a reflected signal that is received by the first receiver Rx1, while the remaining portion passes through the target sensing area SoI and is received by the backward RIS as a transmitted signal. After beam steering by the backward RIS, the beam is focused on the second receiver Rx2.

[0009] The deep learning-based imaging network is used to extract features of the reflection signal received by the first receiver Rx1 and the transmission signal received by the second receiver Rx2, obtain corresponding features for characterizing the reflection coefficient and the transmission coefficient, and generate a contour map of the target after feature fusion.

[0010] Furthermore, the forward RIS and the backward RIS use a configured radiation mode to control the corresponding beams. For the beam pointing to the target sensing area SoI, the configuration strategy for the forward RIS and the backward RIS radiation mode specifically includes:

[0011] Divide the target sensing area SoI into multiple blocks. For each block, adjust the radiation mode of the forward RIS so that it focuses on each block in turn. With the goal of maximizing the signal power received by the current block, use a greedy algorithm to traverse the states of each reflector unit of the forward RIS to obtain a codebook consisting of the states of each reflector unit of the forward RIS corresponding to the current block, thereby determining the radiation mode of the forward RIS.

[0012] Based on the determined radiation mode of the forward RIS, with the goal of maximizing the power of the transmitted signal received by the second receiver Rx2, a greedy algorithm is used to traverse the states of each reflection unit of the backward RIS to obtain a corresponding codebook composed of the states of each reflection unit of the backward RIS to determine the radiation mode of the backward RIS.

[0013] Furthermore, for the beam directed to the rear RIS area, the configuration strategy for the forward RIS and the rear RIS radiation modes specifically includes:

[0014] Dividing the backward RIS into a plurality of blocks based on the area of ​​the backward RIS; regulating the radiation pattern of the forward RIS so that it is focused on the block corresponding to the backward RIS, and using a greedy algorithm to traverse the states of each reflector unit of the forward RIS with the goal of maximizing the signal power received in the current block, to obtain a codebook consisting of the states of each reflector unit of the forward RIS corresponding to the current block, thereby determining the radiation pattern of the forward RIS;

[0015] Based on the determined radiation mode of the forward RIS, with the goal of maximizing the power of the transmitted signal received by the second receiver Rx2, a greedy algorithm is used to traverse the states of each reflection unit of the backward RIS to obtain a corresponding codebook composed of the states of each reflection unit of the backward RIS to determine the radiation mode of the backward RIS.

[0016] Furthermore, for the radiation beam generated by the random phase control of the RIS, the configuration strategy for the forward RIS and the backward RIS radiation modes specifically includes:

[0017] The radiation mode of the forward RIS is random. Under each random radiation mode of the forward RIS, with the goal of maximizing the power of the transmitted signal received by the second receiver Rx2, a greedy algorithm is used to traverse the states of each reflection unit of the backward RIS to obtain a corresponding codebook composed of the states of each reflection unit of the backward RIS to determine the radiation mode of the backward RIS.

[0018] Furthermore, the length of each block divided within the target sensing area SoI is consistent with the main lobe width of the forward RIS.

[0019] Furthermore, the deep learning-based imaging network includes:

[0020] An encoder, configured to encode the reflection signal and the transmission signal, and generate features representing the reflection coefficient and the transmission coefficient respectively;

[0021] The fusion module is used to fuse the features generated by the encoder using an external attention mechanism to obtain enhanced features that focus on the target contour;

[0022] The decoder is used to decode the enhanced features to obtain a contour map of the target.

[0023] Furthermore, the encoder and the decoder are Res-CNN.

[0024] Furthermore, the central operating frequency band of the transmitter Tx is 5.8 GHz.

[0025] According to a second aspect of the present invention, a dual RIS-assisted radio complementary mode imaging method is provided, comprising: using the imaging system described in any one of the first aspects to perform contour imaging of a target in a target perception area SoI to obtain a contour map of the target.

[0026] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0027] (1) The dual-RIS assisted radio complementary mode imaging system of the present invention establishes two radio receiving modes: reflection mode and transmission mode by placing a forward RIS and a backward RIS on both sides of the target sensing area SoI. The target contour is reconstructed based on the two complementary modes to improve the quality of the reconstructed image. Specifically, the forward RIS and the backward RIS are used as electromagnetic control means. The first receiver Rx1 receives the reflection signal of the target object, and the second receiver Rx2 receives the transmission signal passing through the target sensing area SoI, and the reflection signal and the transmission signal of the target object are obtained at the same time. The reflection signal and the transmission signal are respectively extracted to obtain the corresponding features representing the reflection coefficient and the transmission coefficient. After fusion, the target contour is reconstructed. The present invention combines the two forms of electromagnetic signals of reflection and transmission to achieve the effect of complementary imaging. Compared with the existing method of reconstructing images based on the reflection coefficient alone, the present invention reconstructs the target contour based on the reflection and transmission coefficient features of the target object. This method of imaging assisted by the forward and backward RIS fully utilizes the information of the target sensing area SoI beam, greatly improving the quality of the generated image.

[0028] (2) Furthermore, the present invention adopts different scanning strategies for the beam focused on the target sensing area SoI, the beam focused on the backward RIS area, and the radiation beam generated by the random phase control of the previous RIS, respectively, to enhance the reflected signal and the transmitted signal. Specifically, by regulating the radiation mode of the forward RIS, the signal power of the beam focused on different blocks of the target sensing area SoI is maximized, so as to mainly enhance the reflected signal; by regulating the radiation mode of the forward RIS, it is focused on the block corresponding to the backward RIS, so as to mainly enhance the transmitted signal; and, for the radiation beam generated by the random phase control of the previous RIS, the radiation beam is enhanced. At the same time, for the backward RIS, the radiation mode of the backward RIS is regulated based on the electric field distribution of each forward RIS, so as to ultimately maximize the electric field strength of the receiver Rx2, and respectively achieve the enhancement of the reflected signal and the transmitted signal under each strategy. In this way, the scanning of the beam focused on the target sensing area SoI, the beam focused on the backward RIS area, and / or the radiation beam area generated by the random phase control of the previous RIS is covered, so that the spatial resolution of the system is improved. Moreover, the spatial diversity is increased by obtaining RF data (reflection and transmission signals) with diverse patterns under different scanning strategies; while ensuring imaging accuracy, a variety of electromagnetic data can be obtained with fewer sampling times.

[0029] (3) Furthermore, the imaging network based on deep learning designed in the present invention introduces an external attention mechanism into RF data imaging to fuse the features corresponding to the reflection signal and the transmission signal using the external attention mechanism, thereby achieving enhanced features that focus more on the target contour. Compared with the existing method of directly using the codec network for imaging, the present invention combines the external attention mechanism to improve the recognition of data features, can achieve super-resolution imaging results, and improve the robustness and effect of imaging.

[0030] (4) Preferably, the length of each block divided within the target sensing area SoI is consistent with the main lobe width of the forward RIS to maximize the power of the received signal.

[0031] In summary, the present invention combines the perception signal reflection mode signal and the transmission mode signal to achieve imaging. This operation alleviates the signal instability when imaging using only the reflection signal. The hybrid mode dual RIS system of the present invention demonstrates the synergistic and complementary value of this multi-modal combination. In addition, RIS, as a large-aperture antenna in the passive relay "antenna" array, can help Rx achieve a wider field of view and precise perception. Compared to deploying large-scale transceiver equipment, deploying multiple RIS can achieve higher spatial diversity characteristics, realizing an economical and high-freedom imaging method, making the system of the present invention exhibit a high degree of freedom configuration characteristics similar to that of a dual-static radar. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Schematic diagram of an imaging system in a dual RIS-assisted radio complementary mode in an embodiment of the present invention.

[0033] Figure 2 FIG. 4 is a schematic diagram of the hardware structure of an imaging system in a dual RIS-assisted radio complementary mode according to an embodiment of the present invention.

[0034] Figure 3 Schematic diagram of three types of scanning strategies in an embodiment of the present invention; wherein, Figure 3 (a)-(c) are the scanning strategies of focusing the beam on the target sensing area SoI, the scanning strategy of pointing the beam to the posterior RIS, and the scanning strategy of the radiation beam generated by the random phase control of the anterior RIS.

[0035] Figure 4 This is a schematic diagram of the deep learning imaging network structure in an embodiment of the present invention; Figure 4 (a)-(c) are the main network structure, external attention mechanism and Res-CNN network structure respectively.

[0036] Figure 5 This is a visualization diagram of data collection in an embodiment of the present invention.

[0037] Figure 6 Schematic diagram of imaging results and data visualization corresponding to different original images in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0039] In the present invention, the terms "first", "second", etc. in the present invention and the accompanying drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0040] Example 1

[0041] like Figure 1 and Figure 2As shown, an embodiment of the present invention provides an imaging system in a dual-RIS-assisted radio complementary mode, including: a transmitter Tx, a first receiver Rx1, a second receiver Rx2, a forward reconfigurable intelligent surface RIS, a rearward reconfigurable intelligent surface RIS, and an imaging network based on deep learning; wherein the forward RIS and the rearward RIS are placed oppositely on both sides of the target perception area SoI.

[0042] The transmitter Tx points to the forward RIS and is used to transmit linear frequency modulation wave signals;

[0043] The forward RIS is used to perform beam steering on the received linear frequency modulation wave signal so that the steered beam passes through the target sensing area SoI. Part of the beam passing through the target sensing area SoI is reflected by the target, forming a reflected signal and received by the first receiver Rx1. The other part passes through the target sensing area SoI and is received as a transmitted signal by the backward RIS. After beam steering by the backward RIS, the steered electromagnetic signal beam is focused on the second receiver Rx2.

[0044] The deep learning-based imaging network is used to extract features of the reflection signal received by the first receiver Rx1 and the transmission signal received by the second receiver Rx2, and obtain corresponding features for characterizing the reflection coefficient and the transmission coefficient. After feature fusion, a contour map of the target is generated.

[0045] In the embodiment of the present invention, the central operating frequency band of the system is 5.8 GHz, which is also the current operating frequency band of indoor Wi-Fi.

[0046] The dual-RIS-assisted radio complementary imaging system of the present invention establishes two radio reception modes: reflection mode and transmission mode, by placing a forward RIS and a backward RIS on opposite sides of the target sensing area (SoI). These two complementary modes jointly reconstruct the target contour to improve the quality of the reconstructed image. Specifically, using the forward and backward RIS as electromagnetic control means, a first receiver Rx1 receives the target object's reflection signal, while a second receiver Rx2 receives the transmission signal that passes through the target sensing area (SoI), simultaneously acquiring both the target object's reflection and transmission signals. Feature extraction is performed on the reflection and transmission signals separately, obtaining corresponding features representing the reflection coefficient and transmission coefficient. These features are then fused to reconstruct the target contour. This invention combines both reflected and transmitted electromagnetic signals to achieve complementary imaging. Compared to existing methods that reconstruct images based solely on reflection coefficients, the present invention reconstructs the target contour based on the target's reflection and transmission coefficient features. This forward and backward RIS-assisted imaging method fully utilizes the information from the SoI beam, significantly improving the quality of the generated image.

[0047] Experiments have also demonstrated that the system of the present invention can achieve higher super-resolution results than existing 5.8 GHz electromagnetic imaging.

[0048] As a further design of the present invention, when the forward RIS and the backward RIS perform beam steering, they call the configured radiation mode to scan and sense the space in turn to perform beam forming.

[0049] In the embodiment of the present invention, three types of scanning strategies are used to configure the radiation mode of the forward RIS and the backward RIS to enhance the reflected signal and the transmitted signal. Specifically:

[0050] The beams controlled by the RIS can be roughly divided into three categories: beams focused on the target sensing area SoI, beams focused on the backward RIS area, and beams generated by the random phase control of the RIS. Correspondingly, the illumination strategies of the three types of beams are as follows: Figure 3 As shown in (a)-(c) in the figure.

[0051] like Figure 3 As shown in (a) of the figure, the first type of beam scanning strategy focuses the beam on the target sensing area (SoI) (primarily for enhancing reflected signals). The target sensing area (SoI) is divided into blocks, each corresponding to a scanning strategy. The radiation pattern of the forward RIS is controlled to focus on each block in sequence. A greedy algorithm is used to traverse the states of each forward RIS reflector (including 0 and 1). The goal is to maximize the signal power received in the current block. A codebook consisting of the states of each forward RIS reflector corresponding to the current block is obtained. The codebook corresponding to each block is used to determine the radiation pattern of the forward RIS.

[0052] For the backward RIS, a control strategy is set based on the electric field distribution of the forward RIS to reshape the backward electromagnetic field, ultimately maximizing the electric field strength at receiver Rx2. Specifically, based on the determined radiation pattern of the forward RIS, a greedy algorithm is used to traverse the states of each reflector unit in the backward RIS. With the goal of maximizing the signal power received by the second receiver Rx2, a corresponding codebook consisting of the states of each reflector unit in the backward RIS is obtained to determine the radiation pattern of the backward RIS.

[0053] like Figure 3As shown in (b), the second type of scanning strategy is a beam scanning strategy that points the beam toward the backward RIS (primarily for enhancing the transmitted signal). Based on the area of ​​the backward RIS, the backward RIS is divided into multiple blocks. The radiation pattern of the forward RIS is controlled to focus on the block corresponding to the backward RIS. A greedy algorithm is used to traverse the states of each reflector in the forward RIS, with the goal of maximizing the signal power received by the current block. A codebook consisting of the states of each reflector in the forward RIS corresponding to the current block is obtained. The codebook corresponding to each block is used to determine the radiation pattern of the forward RIS. Based on the determined radiation pattern of the forward RIS, a greedy algorithm is used to traverse the states of each reflector in the backward RIS, with the goal of maximizing the signal power received by the second receiver Rx2. A codebook consisting of the states of each reflector in the backward RIS is obtained to determine the radiation pattern of the backward RIS.

[0054] like Figure 3 As shown in (c), the third type of scanning strategy is the radiation beam scanning strategy generated by the random phase control of the RIS. The forward RIS radiation pattern is random. Under each random forward RIS radiation pattern, a greedy algorithm is used to traverse the states of each reflector unit in the backward RIS. The goal is to maximize the signal power received by the second receiver Rx2. The corresponding codebook consisting of the states of each reflector unit in the backward RIS is obtained to determine the radiation pattern of the backward RIS.

[0055] In the embodiment of the present invention, the target sensing area SoI is divided into 35 blocks, each block corresponds to a scanning strategy, a total of 35 types, each focus point of the beam focus area is as follows: Figure 3 As shown in (a) in the figure. Each beam points to each block in the sensing area. The block size is 0.3m*0.3m, which just corresponds to the main lobe width of the forward RIS to maximize the power of the received signal. The second scanning strategy has a total of 25 types, and the spatial beam focus points are as follows: Figure 3 As shown in (b), the size of the RIS used in the embodiment of the present invention is 1.4m*1.4m. The third is the scanning strategy for 10 random radiations, such as Figure 3As shown in (c) in the figure. A greedy traversal algorithm is used to realize the actual acquisition of the beam. In an embodiment of the present invention, under three categories of 70 scanning modes, the system can complete the acquisition of the corresponding RF data (reflection signal and transmission signal) under each scanning mode. For one cycle (70 types of sequential calls and traversals constitute one cycle), the acquisition time is controlled within 1s. By collecting the corresponding RF data at different times, continuous real-time reconstruction of the target can be achieved. In an embodiment of the present invention, a codebook of 70 beam scans is designed. This beam joint control and the obtained RIS control scheme not only ensure the reliability of the collected samples, but also ensure the accurate acquisition of radio data with less perception sampling. In an embodiment of the present invention, the label corresponding to the data collected in each cycle is the image contour of the target in the perception area.

[0056] In other words, this scanning method not only takes into account the energy gain effect brought by beam focusing, but also improves the system's spatial resolution by different scanning areas. Specifically, by regulating the radiation pattern of the forward RIS, the signal power of the beam focused on different blocks of the target sensing area (SoI) is maximized, thereby primarily enhancing the reflected signal. By regulating the radiation pattern of the forward RIS, it is focused on the corresponding blocks of the backward RIS, thereby primarily enhancing the transmitted signal. Furthermore, the radiation beam generated by random phase control of the forward RIS is enhanced. Simultaneously, for the backward RIS, the electromagnetic field of the backward RIS is reshaped based on the electric field distribution of each forward RIS, ultimately maximizing the electric field strength at receiver Rx2. This enhances the reflected and transmitted signals under each strategy. This encompasses scanning of the beam focused on the target sensing area (SoI), the beam focused on the backward RIS area, and the radiation beam generated by random phase control of the forward RIS, thereby improving the system's spatial resolution. Moreover, the spatial diversity is increased by obtaining RF data (reflection and transmission signals) with diverse patterns under different scanning strategies; while ensuring imaging accuracy, a variety of electromagnetic data can be obtained with fewer sampling times.

[0057] As a further design of the present invention, Figure 4 As shown, the deep learning-based imaging network in the embodiment of the present invention includes an encoder, a fusion module and a decoder.

[0058] The encoder is used to encode the reflection signal and the transmission signal, and generate features representing the reflection coefficient and the transmission coefficient respectively;

[0059] The fusion module is used to fuse the features generated by the encoder using an external attention mechanism to generate enhanced features that focus more on the target contour;

[0060] The decoder is used to decode the enhanced features to obtain the contour map of the target.

[0061] As a preferred implementation, the encoder and decoder are Res-CNN.

[0062] The deep learning-based imaging network designed in the present invention introduces an external attention mechanism into RF data imaging to fuse the features corresponding to the reflected signal and the transmitted signal using the external attention mechanism, thereby achieving enhanced features that focus more on the target contour. Compared with the existing method of directly using a codec network for imaging, the present invention combines the external attention mechanism to improve the recognition of data features, achieve super-resolution imaging results, and improve the robustness and effect of imaging.

[0063] The imaging effect of the present invention is described below with reference to specific examples.

[0064] like Figure 5 FIG. 1 is a diagram showing visualization of data collection in an embodiment of the present invention. Figure 5 The first to third rows in the figure represent the original image, and the corresponding reflection signal and transmission signal respectively.

[0065] In the implementation of the present invention, different deep learning networks are used to reconstruct the target contour for the acquired reflection signal and transmission signal data, and the target contour is reconstructed under different evaluation indicators (SSIM, IoU, MAE, F β , BEC), the corresponding experimental data are shown in Table 1 below.

[0066] Table 1 Comparison of ablation experiment results

[0067]

[0068] Among them, SA in Table 1 is the self-attention mechanism network, EA-I is QV as the external attention parameter, and PSD2ImageNet is an imaging network based on deep learning that uses K as the external attention parameter in the embodiment of the present invention. SSIM is the structural similarity index, IoU is an evaluation index used to measure the degree of overlap between the predicted border and the true border, MAE is the mean absolute error, and F β BEC is a single score that combines the two indicators of precision and recall and is calculated by harmonic mean.

[0069] It can be seen that the proposed PSD2ImageNet achieves the best performance in the structural similarity index, which is the main evaluation index, and also achieves relatively good performance in other indicators.

[0070] In the embodiment of the present invention, better performance can be obtained for different actual images, such as Figure 6 As shown, Figure 6 The first row in the figure shows the actual image data. The second, third, and fourth rows show the imaging results using the transmission + reflection signal, the reflection signal, and the transmission signal, respectively. It can be seen that the reconstruction in the second row (using the transmission + reflection signal) yields the best imaging results. The bottom two rows show the actual collected electromagnetic signals (reflection and transmission signals).

[0071] Example 2

[0072] An embodiment of the present invention provides an imaging method in a dual-RIS-assisted radio complementary mode, comprising: performing contour imaging of a target in a target sensing area SoI using the dual-RIS-assisted radio complementary mode imaging system of Example 1 to obtain a contour map of the target.

[0073] For the related technical solutions, please refer to the corresponding description in Example 1 and will not be repeated here.

[0074] Overall, this invention proposes a dual-RIS-assisted imaging deployment scheme and system. The combined control of forward and backward RIS enables more precise electromagnetic signal capture. A deep learning network model with an external attention-based encoding and decoding architecture is proposed. By utilizing images to guide the conversion of radio data into target contours, this embodiment achieves optimal wireless imaging in the 5.8 GHz frequency band, achieving super-resolution imaging.

[0075] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A dual RIS-assisted radio complementary imaging system, characterized in that: include: Transmitter Tx, first receiver Rx1, second receiver Rx2, forward RIS and backward RIS placed on both sides of the target perception area SoI, and a deep learning-based imaging network; The transmitter Tx is directed toward the forward RIS and is used to transmit a linear frequency modulation wave signal; The forward RIS is configured to perform beam steering on the received linear frequency modulation wave signal so that the steered beam passes through the target sensing area SoI. A portion of the beam passing through the target sensing area SoI is reflected by the target to form a reflected signal that is received by the first receiver Rx1, while the remaining portion passes through the target sensing area SoI and is received by the backward RIS as a transmitted signal. After beam steering by the backward RIS, the beam is focused on the second receiver Rx2. The deep learning-based imaging network is used to extract features of the reflection signal received by the first receiver Rx1 and the transmission signal received by the second receiver Rx2, obtain corresponding features for characterizing the reflection coefficient and the transmission coefficient, and generate a contour map of the target after feature fusion.

2. The imaging system according to claim 1, wherein: The forward RIS and the backward RIS use a configured radiation mode to control the corresponding beams. For the beam directed to the target sensing area SoI, the configuration strategy for the forward RIS and the backward RIS radiation mode specifically includes: Divide the target sensing area SoI into multiple blocks. For each block, adjust the radiation mode of the forward RIS so that it focuses on each block in turn. With the goal of maximizing the signal power received by the current block, use a greedy algorithm to traverse the states of each reflector unit of the forward RIS to obtain a codebook consisting of the states of each reflector unit of the forward RIS corresponding to the current block, thereby determining the radiation mode of the forward RIS. Based on the determined radiation mode of the forward RIS, with the goal of maximizing the power of the transmitted signal received by the second receiver Rx2, a greedy algorithm is used to traverse the states of each reflection unit of the backward RIS to obtain a corresponding codebook composed of the states of each reflection unit of the backward RIS to determine the radiation mode of the backward RIS.

3. The imaging system according to claim 2, wherein: For a beam directed toward a backward RIS area, a configuration strategy for the forward RIS and backward RIS radiation modes specifically includes: Dividing the backward RIS into a plurality of blocks based on the area of ​​the backward RIS; regulating the radiation pattern of the forward RIS so that it is focused on the block corresponding to the backward RIS, and using a greedy algorithm to traverse the states of each reflector unit of the forward RIS with the goal of maximizing the signal power received in the current block, to obtain a codebook consisting of the states of each reflector unit of the forward RIS corresponding to the current block, thereby determining the radiation pattern of the forward RIS; Based on the determined radiation mode of the forward RIS, with the goal of maximizing the power of the transmitted signal received by the second receiver Rx2, a greedy algorithm is used to traverse the states of each reflection unit of the backward RIS to obtain a corresponding codebook composed of the states of each reflection unit of the backward RIS to determine the radiation mode of the backward RIS.

4. The imaging system according to claim 2 or 3, characterized in that For the radiation beam generated by the random phase control of the forward RIS, the configuration strategy for the radiation mode of the forward RIS and the backward RIS specifically includes: The radiation mode of the forward RIS is random. Under each random radiation mode of the forward RIS, with the goal of maximizing the power of the transmitted signal received by the second receiver Rx2, a greedy algorithm is used to traverse the states of each reflection unit of the backward RIS to obtain a corresponding codebook composed of the states of each reflection unit of the backward RIS to determine the radiation mode of the backward RIS.

5. The imaging system according to claim 2, wherein: The length of each block divided within the target sensing area SoI is consistent with the main lobe width of the forward RIS.

6. The imaging system according to claim 1, wherein: The deep learning-based imaging network includes: An encoder, configured to encode the reflection signal and the transmission signal, and generate features representing the reflection coefficient and the transmission coefficient respectively; The fusion module is used to fuse the features generated by the encoder using an external attention mechanism to obtain enhanced features that focus on the target contour; The decoder is used to decode the enhanced features to obtain a contour map of the target.

7. The imaging system according to claim 6, wherein: The encoder and the decoder are Res-CNN.

8. The imaging system according to claim 1, wherein: The central operating frequency band of the transmitter Tx is 5.8 GHz.

9. A dual RIS-assisted radio complementary imaging method, characterized in that: include: The imaging system according to any one of claims 1 to 8 is used to perform contour imaging on a target in a target perception area SoI to obtain a contour map of the target.

Images