Ultra-wideband radar image enhancement method, device and equipment based on multi-band fusion

Abstract

The application relates to a multi-frequency band fusion-based ultra-wideband radar image enhancement method, device and equipment. A plurality of single-frequency point radar images are obtained by imaging each sub-frequency point in target echo data. The plurality of single-frequency point radar images are respectively accumulated according to a plurality of preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to the plurality of different frequency bands. Finally, the plurality of frequency band radar imaging and complete radar imaging are simultaneously input into a trained image enhancement neural network to obtain an enhanced radar target image. The method can effectively suppress radar image grating sidelobes and clutter and realize high-resolution image enhancement of a target.

Description

Technical Field

[0001] This application relates to the field of radar image enhancement technology, and in particular to an ultra-wideband radar image enhancement method, apparatus and equipment based on multi-band fusion. Background Technology

[0002] Low-frequency ultra-wideband radar electromagnetic waves possess strong penetrating and resolving capabilities, enabling them to penetrate media such as clouds, rain, fog, foliage, ground surfaces, and walls to detect hidden targets. In particular, their ability to acquire target images and scattering information plays a significant role in both military and civilian applications, including human posture estimation, mine clearance, urban warfare, health monitoring, and early warning and identification.

[0003] However, due to the large spectral span of ultra-wideband radar, information from different frequency bands is not fully utilized. For example, low-frequency images have better sidelobe suppression, while high-frequency images have higher resolution. Directly using ultra-wideband data for imaging will couple target image features from different frequency bands, resulting in low image quality. To fully utilize radar data from different frequency bands while ensuring image quality, it is necessary to propose an ultra-wideband radar image enhancement method based on multi-band fusion. Summary of the Invention

[0004] Therefore, it is necessary to provide an ultra-wideband radar image enhancement method, apparatus, and device based on multi-band fusion that can effectively enhance radar images, addressing the aforementioned technical problems.

[0005] A method for enhancing ultra-wideband radar images based on multi-band fusion, the method comprising:

[0006] Acquire target echo data;

[0007] Imaging is performed on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images;

[0008] The multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0009] The multiple frequency band radar images and the complete radar image are simultaneously input into the trained image enhancement neural network to obtain an enhanced radar target image.

[0010] In one embodiment, imaging based on each sub-frequency point in the target echo data includes:

[0011] In the target echo data, the echo signal captured by each sub-frequency point is estimated by least squares to obtain the corresponding single-frequency radar image.

[0012] In one embodiment, the step of accumulating the multiple single-frequency radar images according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands includes:

[0013] According to multiple preset frequency bands, multiple single-frequency radar images corresponding to each frequency band are coherently accumulated to obtain frequency band radar imaging corresponding to each different frequency band.

[0014] The complete radar image is obtained by coherently accumulating all the single-frequency radar images.

[0015] In one embodiment, the target echo data is obtained by detecting the target using an ultra-wideband MIMO step-frequency radar.

[0016] In one embodiment, training the image augmentation neural network includes:

[0017] Based on multiple simulation points, corresponding simulation target echo data are obtained. Based on each simulation target echo data, multiple single-frequency training images of different frequency bands and a complete training image are generated for each simulation point.

[0018] Based on each simulation point, a corresponding point target truth value label is randomly generated. The point target truth value label is convolved with the point spread function to obtain the truth value image of each simulation point.

[0019] The image enhancement neural network is trained using multiple single-frequency training images generated corresponding to a simulation point and the complete training image as a set of training data to obtain the predicted enhanced image.

[0020] The loss function is calculated based on the predicted enhanced image and the corresponding ground truth image. The parameters in the image enhancement neural network are adjusted according to the calculation result until the calculation result of the loss function converges, thus obtaining the trained image enhancement neural network.

[0021] In one embodiment, the image enhancement neural network is a deep convolutional neural network.

[0022] This application also provides an ultra-wideband radar image enhancement device based on multi-band fusion, the device comprising:

[0023] The target echo data acquisition module is used to acquire target echo data;

[0024] The single-frequency radar image acquisition module is used to perform imaging based on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images.

[0025] The single-frequency radar image accumulation module accumulates the multiple single-frequency radar images according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0026] The radar image enhancement module is used to simultaneously input the radar images of the multiple frequency bands and the complete radar image into a trained image enhancement neural network to obtain an enhanced radar target image.

[0027] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps:

[0028] Acquire target echo data;

[0029] Imaging is performed on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images;

[0030] The multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0031] The multiple frequency band radar images and the complete radar image are simultaneously input into the trained image enhancement neural network to obtain an enhanced radar target image.

[0032] A computer-readable storage medium having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0033] Acquire target echo data;

[0034] Imaging is performed on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images;

[0035] The multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0036] The multiple frequency band radar images and the complete radar image are simultaneously input into the trained image enhancement neural network to obtain an enhanced radar target image.

[0037] The aforementioned ultra-wideband radar image enhancement method, apparatus, and device based on multi-band fusion obtains multiple single-frequency radar images by imaging each sub-frequency point in the target echo data. These single-frequency radar images are then accumulated according to multiple preset frequency bands to obtain frequency band radar images and a complete radar image corresponding to multiple different frequency bands. Finally, the multi-frequency band radar images and the complete radar image are simultaneously input into a trained image enhancement neural network to obtain an enhanced radar target image. This method effectively suppresses radar image sidelobes and clutter while achieving high-resolution image enhancement of the target. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating an ultra-wideband radar image enhancement method based on multi-band fusion in one embodiment.

[0039] Figure 2 This is a flowchart illustrating an image enhancement neural network training method in one embodiment;

[0040] Figure 3 This is a schematic diagram illustrating the overall process of image enhancement implemented by a neural network in one embodiment.

[0041] Figure 4 This is a structural block diagram of an ultra-wideband radar image enhancement device based on multi-band fusion in one embodiment;

[0042] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0044] like Figure 1 As shown, this application provides an ultra-wideband radar image enhancement method based on multi-band fusion, including the following steps:

[0045] Step S100: Obtain target echo data.

[0046] Step S110: Image is performed based on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images.

[0047] Step S120: Multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0048] Step S130: Input the radar images from multiple frequency bands and the complete radar image simultaneously into the trained image enhancement neural network to obtain the enhanced radar target image.

[0049] This method fully utilizes the characteristic information of targets in different frequency bands of ultra-wideband radar, which can effectively suppress radar image sidelobes and clutter, and achieve high-resolution image enhancement of targets.

[0050] In step S100, the target echo data is obtained by the ultra-wideband MIMO step-frequency radar detecting the target.

[0051] Specifically, ultra-wideband radar target echo data can be represented as:

[0052]

[0053] In formula (1), m represents the number of array elements after the ultra-wideband MIMO radar antenna is equivalent, n represents the number of discretized grids of the imaging scene, and k = 2πf i / c, where c represents the speed of light, f i =f0+Δf·n i n i =0···N f -1, N f f0 represents the total number of frequency points and f0 represents the carrier frequency. Represents the discretized echo matrix. The scattering coefficient matrix represents the imaging grid.

[0054] Among them, the transmitted echo distance history R T and the received echo distance history R R for:

[0055]

[0056] In formula (2), the transmitting antenna Tx and the receiving antenna Rx are located at (x... T ,y T ,z T ) and (x R ,y R ,z R The array consists of multiple antenna elements, and the coordinates of the scattering coefficient of the imaging grid are (x, y, z).

[0057] In step S110, imaging based on each sub-frequency point in the target echo data includes: performing least squares estimation on the echo signal captured by each sub-frequency point in the target echo data to obtain the corresponding single-frequency radar image.

[0058] Specifically, for snapshot data at a single frequency point, formula (1) can be written in matrix form, i.e.:

[0059] s=Ax (3)

[0060] In formula (3), s and x represent the echo vector and the scattering coefficient vector of the imaging scene grid at a single frequency point, respectively. It is the observation matrix, defined as:

[0061]

[0062] In this embodiment, by using the least squares method to perform minimum variance unbiased estimation of the snapshot data at a single frequency point, we can obtain:

[0063] x=(A T A) -1 A T s (5)

[0064] In step S120, the multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands. This includes: according to multiple preset frequency bands, coherently accumulating multiple single-frequency radar images corresponding to each frequency band to obtain frequency band radar imaging corresponding to each different frequency band, and coherently accumulating all single-frequency radar images to obtain complete radar imaging.

[0065] Coherent accumulation is expressed as:

[0066]

[0067] In this embodiment, the frequency band range can be customized according to the actual situation.

[0068] In step S130, the radar images corresponding to multiple different frequency bands obtained after processing the target echo data, as well as the complete radar image, are input into the trained image enhancement neural network to obtain the enhanced radar target image.

[0069] In this embodiment, as Figure 2As shown, a method for training an image enhancement neural network is also provided, specifically including: first, obtaining corresponding simulated target echo data based on multiple simulated points; generating multiple single-frequency training images and a complete training image corresponding to each simulated point in different frequency bands based on the simulated target echo data; randomly generating corresponding point target ground value labels based on each simulated point; convolving the point target ground value labels with a point spread function to obtain the ground value image corresponding to each simulated point; using multiple single-frequency training images and the complete training image generated for a simulated point as a set of training data to train the image enhancement neural network to obtain a predicted enhanced image; calculating a loss function based on the predicted enhanced image and the corresponding ground value image; adjusting the parameters in the image enhancement neural network based on the calculation result until the calculation result of the loss function converges, thus obtaining the trained image enhancement neural network. Since low-frequency subband radar images have better grating sidelobe suppression effects, and high-frequency subband radar images have high resolution, the powerful feature extraction capabilities of deep neural networks are more suitable for describing the function mapping relationship between image domains, making it easier to achieve radar image quality enhancement.

[0070] In this embodiment, the image enhancement neural network is a deep convolutional neural network, and its network structure can be as follows: Figure 3 As shown. Specifically, deep convolutional neural networks can employ various methods, including but not limited to: convolutional neural networks (CNN), recurrent neural networks (RNN), deep belief networks (DBN), deep autoencoders, and generative adversarial networks (GAN).

[0071] The aforementioned ultra-wideband radar image enhancement method based on multi-band fusion primarily utilizes the powerful fitting ability of deep neural networks. It takes radar images from multiple sub-bands and the entire frequency band as input, and ground truth values ​​as training labels, to achieve end-to-end enhancement from low-resolution to high-resolution radar images. This method fully leverages target feature information from different frequency bands of ultra-wideband radar. In the actual image enhancement process, it effectively suppresses radar image sidelobes and clutter, while also improving radar image resolution, achieving high-resolution radar image enhancement. This provides high-quality radar images for subsequent abstract applications such as detection, recognition, and image interpretation, demonstrating significant application value and practical significance.

[0072] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0073] In one embodiment, such as Figure 4 As shown, an ultra-wideband radar image enhancement device based on multi-band fusion is provided, comprising: a target echo data acquisition module 200, a single-frequency radar image acquisition module 210, a single-frequency radar image accumulation module 220, and a radar image enhancement module 230, wherein:

[0074] The target echo data acquisition module 200 is used to acquire target echo data;

[0075] The single-frequency radar image acquisition module 210 is used to perform imaging based on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images.

[0076] The single-frequency radar image accumulation module 220 accumulates the multiple single-frequency radar images according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0077] The radar image enhancement module 230 is used to simultaneously input the multiple frequency band radar images and the complete radar images into a trained image enhancement neural network to obtain an enhanced radar target image.

[0078] Specific limitations regarding the multi-band fusion-based ultra-wideband radar image enhancement device can be found in the limitations of the multi-band fusion-based ultra-wideband radar image enhancement method described above, and will not be repeated here. Each module in the aforementioned multi-band fusion-based ultra-wideband radar image enhancement device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0079] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements an ultra-wideband radar image enhancement method based on multi-band fusion. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0080] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0081] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0082] Acquire target echo data;

[0083] Imaging is performed on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images;

[0084] The multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0085] The multiple frequency band radar images and the complete radar image are simultaneously input into the trained image enhancement neural network to obtain an enhanced radar target image.

[0086] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0087] Acquire target echo data;

[0088] Imaging is performed on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images;

[0089] The multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands.

[0090] The multiple frequency band radar images and the complete radar image are simultaneously input into the trained image enhancement neural network to obtain an enhanced radar target image.

[0091] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for enhancing ultra-wideband radar images based on multi-band fusion, characterized in that, The method includes: Acquire target echo data; Imaging is performed on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images; The multiple single-frequency radar images are accumulated according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands. Specifically, according to multiple preset frequency bands, multiple single-frequency radar images corresponding to each frequency band are coherently accumulated to obtain frequency band radar imaging corresponding to each different frequency band, and the complete radar imaging is obtained by coherently accumulating all the single-frequency radar images. The multiple frequency band radar images and the complete radar image are simultaneously input into the trained image enhancement neural network to obtain an enhanced radar target image; Training the image augmentation neural network includes: Based on multiple simulation points, corresponding simulation target echo data are obtained. Based on each simulation target echo data, multiple single-frequency training images of different frequency bands and a complete training image are generated for each simulation point. Based on each simulation point, a corresponding point target truth value label is randomly generated. The point target truth value label is convolved with the point spread function to obtain the truth value image of each simulation point. The image enhancement neural network is trained using multiple single-frequency training images generated corresponding to a simulation point and the complete training image as a set of training data to obtain the predicted enhanced image. The loss function is calculated based on the predicted enhanced image and the corresponding ground truth image. The parameters in the image enhancement neural network are adjusted according to the calculation result until the calculation result of the loss function converges, thus obtaining the trained image enhancement neural network.

2. The ultra-wideband radar image enhancement method based on multi-band fusion according to claim 1, characterized in that, The imaging based on each sub-frequency point in the target echo data includes: In the target echo data, the echo signal captured by each sub-frequency point is estimated by least squares to obtain the corresponding single-frequency radar image.

3. The ultra-wideband radar image enhancement method based on multi-band fusion according to claim 1, characterized in that, The target echo data is obtained by detecting the target using an ultra-wideband MIMO step-frequency radar.

4. The ultra-wideband radar image enhancement method based on multi-band fusion according to claim 3, characterized in that, The image enhancement neural network is a deep convolutional neural network.

5. A multi-band fusion-based ultra-wideband radar image enhancement device, characterized in that, The device implements the ultra-wideband radar image enhancement method based on multi-band fusion as described in any one of claims 1 to 4, including: The target echo data acquisition module is used to acquire target echo data; The single-frequency radar image acquisition module is used to perform imaging based on each sub-frequency point in the target echo data to obtain multiple single-frequency radar images. The single-frequency radar image accumulation module accumulates the multiple single-frequency radar images according to multiple preset frequency bands to obtain frequency band radar imaging and complete radar imaging corresponding to multiple different frequency bands. The radar image enhancement module is used to simultaneously input the radar images of the multiple frequency bands and the complete radar image into a trained image enhancement neural network to obtain an enhanced radar target image.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

Images