Terahertz rotating target imaging method and device based on non-parametric framework

Abstract

A kind of nonparametric framework-based terahertz rotating target imaging method and device, including detecting rotating target according to the difference of rotating target and stationary target echo time-frequency characteristics;Visual saliency feature map is obtained by using spectrum residual visual saliency detection, and background image and rotating target image in visual saliency feature map are separated;Two rotating target azimuth signals in rotating target image are taken as signal of interest, two signals of interest are phase-unravelled and additional Doppler center is estimated, then compensation function is constructed to compensate phase error of rotating target image, and after refocusing, the rotating target image is obtained, and finally complete high-resolution image is output.The present application avoids time-consuming parameter estimation of traditional parameterized method, and can adaptively consider background clutter, providing technical support for high-resolution rotating target imaging.

Description

Technical Field

[0001] This invention relates to the fields of radar signal processing technology and radar imaging technology, and in particular to a terahertz rotating target imaging method and apparatus based on a nonparametric frame. Background Technology

[0002] Synthetic Aperture Radar (SAR) can acquire high-quality images of stationary areas at almost any time, under various weather and lighting conditions. The demands for dynamic situational awareness and rotating target detection have driven the development of SAR, leading to Video Synthetic Aperture Radar (ViSAR). ViSAR extends the temporal resolution of images within a given imaging resolution, thus enabling the monitoring of regions of interest through image sequences. For high-resolution radar imaging scenarios with urgent applications, ViSAR is widely used in dynamic urban reconnaissance and surveillance, as well as for indicating slowly rotating targets.

[0003] Target / component-level motion models can generally be divided into two categories. The first category is defined as linear motion at the target level, such as uniform velocity or uniform acceleration. The second category refers to curvilinear micro-motion at the target / component level, such as rotation, with representative structures including the search antenna of a cruise ship and the rotor blades of a hovering helicopter. Identifying these rotating components is beneficial for refined target recognition. However, while rotation can expand the dimension of target features, it can also generate time-varying Doppler frequencies with periodic or nonlinear modulation, thereby disrupting the signal foundation of traditional imaging algorithms. Under certain specific conditions, such rotation may trigger active interference or electromagnetic countermeasures attacks. Therefore, micro-motion target indication (MMTI) and imaging (MMT imaging, MMTIm) for rotating targets are of urgent significance. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention proposes a terahertz rotating target imaging method and device based on a non-parametric frame.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] On one hand, the present invention provides a terahertz rotating target imaging method based on a nonparametric frame, comprising:

[0007] The original linear frequency modulated echo signal of the terahertz radar system is divided into apertures to obtain sub-aperture echo signals;

[0008] A nonparametric time-frequency transformation is performed on the sub-aperture echo signal, and the rotating target is detected based on the difference in the time-frequency characteristics of the echoes of rotating and stationary targets.

[0009] After confirming the existence of a rotating target, the original sub-aperture echo signal containing the defocused rotating target is subjected to visual saliency detection based on spectral residuals to obtain a visual saliency feature map, and the background image and the rotating target image in the visual saliency feature map are separated.

[0010] Two rotating target orientation signals in the rotating target image are taken as signals of interest. The two signals of interest are phase unwrapped. Based on the phase unwrapping results, a phase error compensation function and a range cell migration compensation function are constructed.

[0011] The extra Doppler center is estimated based on the relative relationship of the azimuth spectra of two signals of interest, and a compensation function for the extra Doppler center is constructed.

[0012] Phase error compensation is performed on the rotating target image based on the phase error compensation function, the range cell migration compensation function, and the compensation function of the extra Doppler center to obtain the refocused rotating target image.

[0013] The refocused rotated target image is stitched together with the residual background image to output a complete high-resolution image.

[0014] On the other hand, a terahertz rotating target imaging device based on a nonparametric frame is provided, comprising:

[0015] The first module is used to divide the original linear frequency modulated echo signal of the terahertz radar system into apertures to obtain sub-aperture echo signals.

[0016] The second module is used to perform nonparametric time-frequency transformation on the sub-aperture echo signal and detect the rotating target based on the difference in time-frequency characteristics of the echoes of the rotating target and the stationary target.

[0017] The third module is used to obtain a visual saliency map by performing spectral residual visual saliency detection on the original sub-aperture echo signal containing the defocused rotating target after confirming the existence of the rotating target, and to separate the background image and the rotating target image in the visual saliency map.

[0018] The fourth module is used to take the two rotating target orientation signals in the rotating target image as signals of interest, perform phase unwrapping on the two signals of interest, and construct a phase error compensation function and a range cell migration compensation function based on the phase unwrapping results.

[0019] The fifth module is used to estimate the extra Doppler center based on the relative relationship of the azimuth spectra of two signals of interest, and to construct a compensation function for the extra Doppler center;

[0020] The sixth module is used to perform phase error compensation on the rotating target image based on the phase error compensation function, the range cell migration compensation function, and the compensation function of the extra Doppler center, so as to obtain the refocused rotating target image.

[0021] The seventh module is used to stitch the refocused rotated target image with the residual background image to output a complete high-resolution image.

[0022] This invention can adaptively estimate the phase error and additional Doppler center from a rotating target without requiring additional consideration of the motion model of the unknown target. It also adaptively considers background clutter interference without increasing system complexity, exhibiting high efficiency and robustness. This method avoids the time-consuming parameter estimation of traditional parameterization methods and adaptively considers background clutter, providing technical support for high-resolution imaging of rotating targets. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0024] Figure 1 A flowchart of a terahertz rotating target imaging method based on a nonparametric frame provided in an embodiment of the present invention;

[0025] Figure 2 The following is a time-frequency imaging result diagram of an embodiment under 5 consecutive frames, wherein (a) is the time-frequency imaging result diagram of the first frame, (b) is the time-frequency imaging result diagram of the second frame, (c) is the time-frequency imaging result diagram of the third frame, (d) is the time-frequency imaging result diagram of the fourth frame, and (e) is the time-frequency imaging result diagram of the fifth frame.

[0026] Figure 3 The image shows the original imaging results in five consecutive frames, as shown in the example. (a) Original imaging result in the first frame, (b) Original imaging result in the second frame, (c) Original imaging result in the third frame, (d) Original imaging result in the fourth frame, and (e) Original imaging result in the fifth frame.

[0027] Figure 4 This is an example of superimposing visual salient feature maps and corresponding original images in 5 consecutive frames, where (a) is the visual salient feature map in 5 consecutive frames, and (b) is the effect image after superimposing the visual salient feature map and the corresponding original image in 5 consecutive frames.

[0028] Figure 5This is a diagram of the defocused target area where the four-sided rotating corner reflector is located in five consecutive frames, according to an embodiment.

[0029] Figure 6 The original phase and unwrapped phase of the signal of interest extracted in the 4th frame in one embodiment are shown in the figure. (a) is the original phase of the four-sided rotating corner reflector 1, (b) is the unwrapped phase of the four-sided rotating corner reflector 1, (c) is the original phase of the four-sided rotating corner reflector 2, and (d) is the unwrapped phase of the four-sided rotating corner reflector 2.

[0030] Figure 7 The image shows the refocusing results of a four-sided rotating corner reflector in five consecutive frames, where (a) is the refocusing result of the four-sided rotating corner reflector in the first frame, (b) is the refocusing result of the four-sided rotating corner reflector in the second frame, (c) is the refocusing result of the four-sided rotating corner reflector in the third frame, (d) is the refocusing result of the four-sided rotating corner reflector in the fourth frame, and (e) is the refocusing result of the four-sided rotating corner reflector in the fifth frame.

[0031] Figure 8 The following is a final video imaging result image corresponding to the original imaging result image of a four-sided rotating corner reflector in 5 consecutive frames, wherein (a) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the first frame, (b) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the second frame, (c) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the third frame, (d) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the fourth frame, and (e) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the fifth frame. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0033] Reference Figure 1 In one embodiment, a terahertz rotating target imaging method based on a nonparametric frame is provided, comprising:

[0034] The original linear frequency modulated echo signal of the terahertz radar system is divided into apertures to obtain sub-aperture echo signals;

[0035] A nonparametric time-frequency transformation is performed on the sub-aperture echo signal, and the rotating target is detected based on the difference in the time-frequency characteristics of the echoes of rotating and stationary targets.

[0036] After confirming the existence of a rotating target, the original sub-aperture echo signal containing the defocused rotating target is subjected to visual saliency detection based on spectral residuals to obtain a visual saliency feature map, and the background image and the rotating target image in the visual saliency feature map are separated.

[0037] Two rotating target orientation signals in the rotating target image are taken as signals of interest. The two signals of interest are phase unwrapped. Based on the phase unwrapping results, a phase error compensation function and a range cell migration compensation function are constructed.

[0038] The extra Doppler center is estimated based on the relative relationship of the azimuth spectra of two signals of interest, and a compensation function for the extra Doppler center is constructed.

[0039] Phase error compensation is performed on the rotating target image based on the phase error compensation function, the range cell migration compensation function, and the compensation function of the extra Doppler center to obtain the refocused rotating target image.

[0040] The refocused rotated target image is stitched together with the residual background image to output a complete high-resolution image.

[0041] In one embodiment, a method for detecting a rotating target based on the difference in time-frequency characteristics of the echoes of a rotating target and a stationary target is provided, including:

[0042] Convert the sub-aperture echo signal to the range Doppler domain;

[0043] The sub-aperture echo signal in the range-Doppler domain is subjected to azimuth compression and inverse Fourier transform, and then converted to the time-frequency domain to obtain the time-frequency result.

[0044] The rotating target is detected in the time-frequency result based on the difference in time-frequency characteristics between the echoes of the rotating target and the stationary target. The time-frequency characteristics of the echo of the rotating target are different from those of the echo of the stationary target. The echo of the rotating target carries sinusoidal modulation, thereby detecting the rotating target.

[0045] In one embodiment, a visual saliency feature map obtained based on spectral residual visual saliency detection is provided as follows:

[0046] S(x)=g(x)*IFFT2[exp(R(f)+P(f))] 2

[0047] In this context, g(x) represents a Gaussian filter, IFFT2 represents the two-dimensional inverse Fourier transform in the image domain, R(f) represents the spectral residual, and P(f) represents the phase spectrum.

[0048] Spectral residual and phase spectrum are both well-known definitions in the field of image processing. Spectral residual refers to the logarithmic spectral domain of an amplitude image and the residual components of that amplitude image after mean filtering. The data of the amplitude image after two-dimensional inverse Fourier transform is complex data. The phase spectrum is the corresponding phase of this complex data.

[0049] The imaging results of rotating targets under traditional imaging methods have special characteristics (they appear as long tails in the image and are off-center from the imaging scene). Video synthetic aperture radar imaging can obtain imaging results of rotating targets in different states, while the imaging results of stationary targets remain relatively stable under small apertures. By comparing the differences between the images, the background image and the rotating target image in the visual salient feature map can be separated.

[0050] In one embodiment, a method for phase unwrapping two signals of interest is provided:

[0051] Let the actual phase distribution of two signals of interest in slow time η be as follows: Where m = 1, 2, These represent the actual phase distributions of the first and second signals of interest at slow time η, respectively.

[0052] Define the unwarp(·) operator:

[0053]

[0054] Where k is an integer not less than 2. It is the phase of the kth and (k-1)th sampling points, η = k / PRF, where PRF is the pulse repetition frequency of the terahertz radar system;

[0055] By unwrapping the phases of two signals of interest, the phase distributions of the unwrapped signals are obtained.

[0056]

[0057] in These represent the phase distributions of the first and second signals of interest after phase unwrapping, respectively.

[0058] In one embodiment, a phase error compensation function for constructing two signals of interest is provided as follows:

[0059]

[0060] In one embodiment, a distance cell migration compensation function for constructing two signals of interest is provided as follows:

[0061]

[0062] Where f τ and f c These are the distance frequency and carrier frequency of the sub-aperture echo signal, respectively.

[0063] In one embodiment, a compensation function for constructing additional Doppler centers for two signals of interest is provided as follows:

[0064] H 2_m (η)=exp(-j2πΔf m η)

[0065] Where Δf m These are the additional Doppler centers of the signals of interest, where Δf1 and Δf2 are the additional Doppler centers of the first and second signals of interest, respectively.

[0066] Δf1=-Δf2=argmax(S1(η)) / 2-argmax(S2(η)) / 2

[0067] S1(η) and S2(η) are the azimuth spectra of the first and second signals of interest, respectively, and argmax(·) represents the position where the maximum value is taken.

[0068] In one embodiment, a refocused rotated target image is provided as follows:

[0069] s final_SoI (τ,η)=IFT η (s rcmc_t (τ,f η )·FT η (H 1_m (η)))

[0070] Among them IFT η and FT η Representing the inverse Fourier transform and Fourier transform along the azimuth dimension, respectively, and the signal after correcting the range cell migration by rotating the target, it is expressed as:

[0071]

[0072] Among them IFT τ and FT τ Representing the inverse Fourier transform and Fourier transform along the distance dimension, respectively, A is the signal amplitude, c is the speed of light, B is the signal bandwidth, and f is the signal amplitude. c It is the radar carrier frequency, f η and These represent the azimuth frequency and the Doppler center frequency, respectively. m (f η Frequency domain representation of the distance history of a rotating target, R Bw represents the closest distance between the rotating target and the trajectory of the moving vehicle equipped with the terahertz system. a (·) indicates an orientation window.

[0073] In one embodiment, the final output is a complete high-resolution image s. final (τ,η), which includes both the moving target and the stationary scene, is represented as:

[0074]

[0075] Where s background P represents the background image. m The distance cell represents the location of the rotating target, and τ and η represent the fast time and slow time in synthetic aperture imaging, respectively.

[0076] This invention has been experimentally verified. Taking a terahertz radar system with a carrier frequency of 216 GHz as an example, the rotating target is two four-sided rotating corner reflectors. The effectiveness of this method is demonstrated by comparing time-frequency images and results before and after focusing. In the experiment, the system carrier frequency was 216 GHz, the bandwidth was 0.9 GHz, the pulse repetition frequency was 16000 Hz, the number of sampling points in each pulse was 8192, the observation time of each frame was 0.2 seconds, the experimental target was placed on an airport 3200 meters away from the radar, the rotation speed was approximately 720° / s, and the rotation radius was approximately 0.32 meters.

[0077] Figure 2 The results of MMTI based on time-frequency imaging are presented across five consecutive frames. These images are presented in logarithmic form, with the dynamic range limited to 40 dB. Compared to the background, the time-frequency representation and intensity of the four rotating corner reflectors are significantly different, allowing for good differentiation and confirming the presence of rotating targets. Figure 3 The original imaging results of the target are shown in these 5 frames. It can be found that the imaging results of the rotated target are out of focus and scattered in the center of the imaging scene.

[0078] Next, taking the first frame of imaging results as an example, visual salient features are applied. Figure 4 The corresponding visual saliency maps and their corresponding original image overlays are shown, where (a) is the visual saliency map over 5 consecutive frames, and (b) is the result of overlaying the visual saliency map and the corresponding original image over 5 consecutive frames. It can be observed that the four-sided rotating corner reflector is effectively detected in the visual saliency map, but stationary warehouses and aircraft also exist. Fortunately, these can be eliminated through the temporal correlation of ViSAR, allowing for rapid identification of the specific range cells where rotating targets exist. Figure 5 It shows the defocused target area that has been captured.

[0079] After completing MMTI, the signal of interest can be extracted. Without loss of generality, taking the fourth frame as an example, Figure 6 The original and unwrapped phases of the two signals of interest (quadrilateral rotating corner reflector 1 and quadrilateral rotating corner reflector 2) are shown. Figure 6 The original phase and unwrapped phase of the signal of interest extracted in the 4th frame in one embodiment are shown in the figure. (a) is the original phase of the four-sided rotating corner reflector 1, (b) is the unwrapped phase of the four-sided rotating corner reflector 1, (c) is the original phase of the four-sided rotating corner reflector 2, and (d) is the unwrapped phase of the four-sided rotating corner reflector 2.

[0080] It is easy to see that they exhibit a clear sinusoidal form and a distinct symmetry. Based on these two signals of interest, the phase error and the additional Doppler center can be calculated, thus completing the refocusing of the rotating target. The final refocusing result is as follows: Figure 7 As shown, (a) is the refocusing result of the four-sided rotating corner reflector in the first frame, (b) is the refocusing result of the four-sided rotating corner reflector in the second frame, (c) is the refocusing result of the four-sided rotating corner reflector in the third frame, (d) is the refocusing result of the four-sided rotating corner reflector in the fourth frame, and (e) is the refocusing result of the four-sided rotating corner reflector in the fifth frame. Finally, the refocused target image is stitched to the residual background to obtain the result shown. Figure 8 As a result, Figure 8 The following is a final video imaging result image corresponding to the original imaging result image of a four-sided rotating corner reflector in 5 consecutive frames, wherein (a) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the first frame, (b) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the second frame, (c) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the third frame, (d) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the fourth frame, and (e) is the final video imaging result image corresponding to the original imaging result image of the four-sided rotating corner reflector in the fifth frame.

[0081] One embodiment provides a terahertz rotating target imaging device based on a non-parametric frame, comprising:

[0082] The first module is used to divide the original linear frequency modulated echo signal of the terahertz radar system into apertures to obtain sub-aperture echo signals.

[0083] The second module is used to perform nonparametric time-frequency transformation on the sub-aperture echo signal and detect the rotating target based on the difference in time-frequency characteristics of the echoes of the rotating target and the stationary target.

[0084] The third module is used to obtain a visual saliency map by performing spectral residual visual saliency detection on the original sub-aperture echo signal containing the defocused rotating target after confirming the existence of the rotating target, and to separate the background image and the rotating target image in the visual saliency map.

[0085] The fourth module is used to take the two rotating target orientation signals in the rotating target image as signals of interest, perform phase unwrapping on the two signals of interest, and construct a phase error compensation function and a range cell migration compensation function based on the phase unwrapping results.

[0086] The fifth module is used to estimate the extra Doppler center based on the relative relationship of the azimuth spectra of two signals of interest, and to construct a compensation function for the extra Doppler center;

[0087] The sixth module is used to perform phase error compensation on the rotating target image based on the phase error compensation function, the range cell migration compensation function, and the compensation function of the extra Doppler center, so as to obtain the refocused rotating target image.

[0088] The seventh module is used to stitch the refocused rotated target image with the residual background image to output a complete high-resolution image.

[0089] The implementation methods of the above modules and the construction of the model can all adopt the methods described in any of the foregoing embodiments, and will not be repeated here.

[0090] On the other hand, the present invention provides a computer device including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the terahertz rotating target imaging method based on a non-parametric frame provided in any of the above embodiments. The computer device may be a server. The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store sample data. The network interface of the computer device is used for communication with external terminals via a network connection.

[0091] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the terahertz rotating target imaging method based on a nonparametric frame provided in any of the above embodiments.

[0092] 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.

[0093] Matters not covered in this invention are common knowledge.

[0094] 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.

[0095] 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.

[0096] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A terahertz rotating target imaging method based on a nonparametric frame, characterized in that, include: The original linear frequency modulated echo signal of the terahertz radar system is divided into apertures to obtain sub-aperture echo signals; A nonparametric time-frequency transformation is performed on the sub-aperture echo signal, and the rotating target is detected based on the difference in the time-frequency characteristics of the echoes of rotating and stationary targets. After confirming the existence of a rotating target, the original sub-aperture echo signal containing the defocused rotating target is subjected to visual saliency detection based on spectral residuals to obtain a visual saliency feature map, and the background image and the rotating target image in the visual saliency feature map are separated. Two rotating target azimuth signals from the rotating target image are taken as signals of interest. Phase unwrapping is performed on these two signals of interest. Based on the phase unwrapping results, a phase error compensation function and a range cell migration compensation function are constructed. The method for phase unwrapping the two signals of interest is as follows: Suppose two signals of interest occur in slow time. The corresponding actual phase distribution is ,in , , These represent the first and second signals of interest in slow time, respectively. The corresponding actual phase distribution; definition Operator: in It is an integer not less than 2. , It is the first The, the Phase of each sampling point ,in PRF The pulse repetition frequency of the terahertz radar system; By unwrapping the phases of two signals of interest, the phase distributions of the unwrapped signals are obtained. , in , These represent the phase distributions of the first and second signals of interest after phase unwrapping, respectively. The phase error compensation function for the two signals of interest is constructed as follows: The distance cell migration compensation function for the two signals of interest is constructed as follows: in and These are the distance frequency and carrier frequency of the sub-aperture echo signal, respectively. The additional Doppler center is estimated based on the relative relationship of the azimuth spectra of the two signals of interest. The compensation function for the additional Doppler center of the two signals of interest is constructed as follows: in It is the additional Doppler center of the signal of interest. , The additional Doppler centers of the first and second signals of interest are respectively: , These are the azimuth spectra of the first and second signals of interest, respectively. Indicates the position of the maximum value; Phase error compensation is performed on the rotating target image based on the phase error compensation function, the range cell migration compensation function, and the compensation function of the extra Doppler center to obtain the refocused rotating target image. The refocused rotated target image is stitched together with the residual background image to output a complete high-resolution image.

2. The terahertz rotating target imaging method based on a nonparametric frame according to claim 1, characterized in that, Detecting rotating targets based on the difference in time-frequency characteristics of echoes from rotating and stationary targets includes: Convert the sub-aperture echo signal to the range Doppler domain; The sub-aperture echo signal in the range-Doppler domain is subjected to azimuth compression and inverse Fourier transform, and then converted to the time-frequency domain to obtain the time-frequency result. The rotating target is detected in the time-frequency result based on the difference in time-frequency characteristics between the echoes of the rotating target and the stationary target. The time-frequency characteristics of the echo of the rotating target are different from those of the echo of the stationary target. The echo of the rotating target carries sinusoidal modulation, thereby detecting the rotating target.

3. The terahertz rotating target imaging method based on a nonparametric frame according to claim 1, characterized in that, The visual saliency feature map obtained based on spectral residual visual saliency detection is as follows: middle Indicates a Gaussian filter. Represents the two-dimensional inverse Fourier transform in the image domain. Represents spectral residuals, Represents the phase spectrum.

4. The terahertz rotating target imaging method based on a nonparametric frame according to claim 1, 2, or 3, characterized in that, The refocused image of the rotated target is as follows: in and Representing the inverse Fourier transform and Fourier transform along the azimuth dimension, respectively, and the signal after correcting the range cell migration by rotating the target, it is expressed as: in and These represent the inverse Fourier transform and the Fourier transform along the distance dimension, respectively. A It is the signal amplitude. c It's the speed of light. B It is the signal bandwidth. It is the radar carrier frequency. and These represent the azimuth frequency and the Doppler center frequency, respectively. Frequency domain representation of the distance history of a rotating target. This indicates the closest distance between the rotating target and the trajectory of the moving vehicle equipped with the terahertz system. Indicates the orientation window.

5. The terahertz rotating target imaging method based on a nonparametric frame according to claim 4, characterized in that, Output complete high-resolution image It includes both moving targets and static scenes, and is represented as: in Represents the background image. This indicates the distance unit where the rotating target is located. and These represent the fast and slow times in synthetic aperture imaging, respectively.

6. A terahertz rotating target imaging device based on a nonparametric frame, used to implement the terahertz rotating target imaging method based on a nonparametric frame as described in claim 1, 2, 3, or 5, characterized in that, include: The first module is used to divide the original linear frequency modulated echo signal of the terahertz radar system into apertures to obtain sub-aperture echo signals. The second module is used to perform nonparametric time-frequency transformation on the sub-aperture echo signal and detect the rotating target based on the difference in time-frequency characteristics of the echoes of the rotating target and the stationary target. The third module is used to obtain a visual saliency map by performing spectral residual visual saliency detection on the original sub-aperture echo signal containing the defocused rotating target after confirming the existence of the rotating target, and to separate the background image and the rotating target image in the visual saliency map. The fourth module is used to take the two rotating target orientation signals in the rotating target image as signals of interest, perform phase unwrapping on the two signals of interest, and construct a phase error compensation function and a range cell migration compensation function based on the phase unwrapping results. The fifth module is used to estimate the extra Doppler center based on the relative relationship of the azimuth spectra of two signals of interest, and to construct a compensation function for the extra Doppler center; The sixth module is used to perform phase error compensation on the rotating target image based on the phase error compensation function, the range cell migration compensation function, and the compensation function of the extra Doppler center, so as to obtain the refocused rotating target image. The seventh module is used to stitch the refocused rotated target image with the residual background image to output a complete high-resolution image.

Images