Near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and space adaptive time-frequency window function

Abstract

The embodiment of the invention provides a near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and a space adaptive time-frequency window function, and is applied to the technical field of radar signal processing. The method comprises the following steps: acquiring an MIMO radar echo signal of a target scene, and generating a wavenumber domain image based on the MIMO radar echo signal; performing phase gradient self-focusing processing based on the wave number domain image to obtain a phase correction image; performing spatial adaptive division based on the phase correction image to obtain a plurality of sub-block images; window function matching is carried out on each sub-block image, a target window function corresponding to each sub-block image is determined, windowing is carried out on the sub-block images based on the target window functions, and windowed sub-block images are obtained; fusing the plurality of windowed sub-block images based on a preset fusion strategy to obtain a fused image; and performing secondary phase gradient self-focusing processing on the fused image to obtain a target imaging result. The technical effect of improving the definition of the imaging result is achieved.

Description

Technical Field

[0001] This application relates to the field of radar signal processing technology, and in particular to a near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function. Background Technology

[0002] Near-field MIMO radar is a novel radar technology that combines multiple input / output (MIMO) with near-field radar imaging. Its core value lies in resolving the contradiction between high resolution and compact structure in traditional radar, and it is now widely used in various fields. In near-field MIMO radar imaging, sidelobes exist as a major interference factor. Therefore, to improve the accuracy and clarity of imaging, effective suppression of sidelobes is necessary.

[0003] In existing technologies, the main approach to sidelobe suppression in near-field MIMO radar imaging is to utilize window functions. Essentially, this involves non-uniformly weighting the echo signal from the near-field MIMO radar to reduce the abrupt changes in time-domain truncation, thereby reducing frequency-domain spectral leakage and achieving sidelobe suppression.

[0004] Because existing technologies mainly achieve sidelobe suppression through fixed window functions, they suffer from low image clarity. Summary of the Invention

[0005] This application provides a near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function to achieve the technical effect of improving the clarity of imaging results.

[0006] In a first aspect, embodiments of this application provide a near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function, including:

[0007] Acquire MIMO radar echo signals of the target scene and generate wavenumber domain images based on the MIMO radar echo signals;

[0008] Phase gradient autofocusing is performed on wavenumber domain images to obtain phase-corrected images; phase gradient autofocusing is used to eliminate phase deviations in wavenumber domain images and improve the clarity of image boundaries.

[0009] Based on the phase-corrected image, spatial adaptive partitioning is performed to obtain multiple sub-block images;

[0010] For each sub-block image, window function matching is performed to determine the target window function corresponding to each sub-block image. The sub-block image is then windowed based on the target window function to obtain a windowed sub-block image. The window function is used to perform weighted calculations on the sub-block image to improve the clarity of the sub-block image boundaries.

[0011] Multiple windowed block images are fused based on a preset fusion strategy to obtain a fused image;

[0012] The fused image is subjected to secondary phase gradient autofocus processing to obtain the target imaging result.

[0013] Optionally, in one possible implementation, phase gradient autofocusing is performed based on the wavenumber domain image to obtain a phase-corrected image, including:

[0014] For each fixed range direction in the wavenumber domain image, the first two-dimensional imaging sub-image corresponding to the fixed range direction is extracted from the wavenumber domain image;

[0015] Based on the first two-dimensional imaging sub-image, block processing is performed to obtain multiple overlapping blocks;

[0016] Phase gradient calculation is performed based on multiple overlapping blocks to obtain the global gradient map corresponding to the first two-dimensional imaging sub-map;

[0017] The phase error is calculated based on the global gradient map, and compensation calculation is performed based on the phase error to obtain the phase correction sub-map corresponding to the first two-dimensional imaging sub-map.

[0018] A phase-corrected image is generated based on multiple phase-corrected sub-images corresponding to fixed distances.

[0019] Optionally, in one possible implementation, the first two-dimensional imaging sub-image is divided into multiple overlapping blocks, including:

[0020] The first two-dimensional imaging sub-image is divided into multiple overlapping blocks; wherein, a single overlapping region occupies half of the overlapping block;

[0021] The center of each overlapping block is moved to the pixel with the highest signal strength in the overlapping block, resulting in multiple shifted overlapping blocks.

[0022] Optionally, in one possible implementation, phase gradient calculation is performed based on multiple overlapping blocks to obtain a global gradient map corresponding to the first two-dimensional imaging sub-map, including:

[0023] Based on each shifted overlapping block, the center row and center column of the shifted overlapping block are obtained, and windowing calculation is performed based on the center row and center column to obtain the windowed overlapping block;

[0024] Phase gradient calculation is performed on each windowed overlapping block to obtain the phase gradient value of each windowed overlapping block;

[0025] The phase gradient values ​​of each windowed overlapping block are interpolated and mapped to the position of the windowed overlapping block in the first two-dimensional imaging sub-image to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image.

[0026] Optionally, in one possible implementation, spatial adaptive partitioning is performed based on the phase-corrected image to obtain multiple sub-block images, including:

[0027] For each fixed distance direction of the phase-corrected image, the second two-dimensional imaging sub-image corresponding to the phase-corrected image is extracted from the phase-corrected image;

[0028] The second two-dimensional imaging sub-image is uniformly divided into multiple basic sub-block images of the same size;

[0029] For each basic sub-block image, the spectral width and energy values ​​of the basic sub-block image are calculated; where the spectral width represents the signal frequency dispersion of the basic sub-block image; and the energy value represents the target reflection intensity of the basic sub-block image, where the target reflection intensity refers to the core target of MIMO radar imaging.

[0030] Based on the spectral width and energy values, each basic sub-block image is adaptively refined to obtain multiple sub-block images.

[0031] Optionally, in one possible implementation, for each basic sub-block image, the spectral width and energy values ​​of the basic sub-block image are calculated, including:

[0032] Set a sliding window of a preset size in each basic sub-block image;

[0033] Perform a two-dimensional Fourier transform on each sliding window to obtain the wavenumber domain signal corresponding to the sliding window;

[0034] The spectral width and energy values ​​of each basic sub-block image are calculated based on the wavenumber domain signal.

[0035] Optionally, in one possible implementation, each basic sub-block image is adaptively refined based on spectral width and energy values ​​to obtain multiple sub-block images, including:

[0036] Based on the spectral width and energy values ​​of multiple basic sub-block images, the energy threshold and spectral width threshold of the second two-dimensional imaging sub-image are calculated.

[0037] For each basic sub-block image, when the spectral width value is greater than the spectral width threshold or the energy value is greater than the energy threshold, the basic sub-block image is divided into multiple first-level refined sub-blocks;

[0038] For each first-level refinement sub-block, if the spectral width value of the first-level refinement sub-block is greater than the spectral width threshold, or the energy value of the first-level refinement sub-block is greater than the energy threshold, the first-level refinement sub-block is divided into multiple second-level refinement sub-blocks.

[0039] Multiple sub-block images are obtained based on multiple basic sub-block images, multiple first-level thinning sub-blocks, and multiple second-level thinning sub-blocks.

[0040] Optionally, in one possible implementation, multiple windowed block images are fused based on a preset fusion strategy to obtain a fused image, including:

[0041] Each windowed sub-block image is expanded outwards to obtain an expanded sub-block image; there are overlapping areas between the expanded sub-block images.

[0042] For every two adjacent dilated sub-block images, determine two target window functions corresponding to the two dilated sub-block images;

[0043] Based on the preset fusion strategy and the function types corresponding to the two target window functions, the target fusion method of two adjacent expanded sub-block images is determined.

[0044] Based on the fusion method corresponding to each adjacent expanded sub-block image, fusion is performed on each overlapping region to obtain a fused image.

[0045] Secondly, embodiments of this application provide a near-field MIMO radar sidelobe suppression device based on phase gradient self-focusing and spatial adaptive time-frequency window function, comprising:

[0046] The acquisition module is used to acquire MIMO radar echo signals of the target scene and generate wavenumber domain images based on the MIMO radar echo signals.

[0047] The first processing module is used to perform phase gradient autofocusing processing on the wavenumber domain image to obtain a phase-corrected image; phase gradient autofocusing is used to eliminate phase deviation in the wavenumber domain image and improve the clarity of the image boundary.

[0048] The second processing module is used to perform spatial adaptive partitioning based on the phase-corrected image to obtain multiple sub-block images;

[0049] The third processing module is used to perform window function matching for each sub-block image, determine the target window function corresponding to each sub-block image, and window the sub-block image based on the target window function to obtain windowed sub-block images; the window function is used to perform weighted calculation on the sub-block images to improve the clarity of the sub-block image boundaries;

[0050] The fourth processing module is used to fuse multiple windowed block images based on a preset fusion strategy to obtain a fused image;

[0051] The fifth processing module is used to perform secondary phase gradient autofocus processing on the fused image to obtain the target imaging result.

[0052] Optionally, in one possible implementation, the first processing module is further configured to:

[0053] For each fixed range direction in the wavenumber domain image, the first two-dimensional imaging sub-image corresponding to the fixed range direction is extracted from the wavenumber domain image;

[0054] Based on the first two-dimensional imaging sub-image, block processing is performed to obtain multiple overlapping blocks;

[0055] Phase gradient calculation is performed based on multiple overlapping blocks to obtain the global gradient map corresponding to the first two-dimensional imaging sub-map;

[0056] The phase error is calculated based on the global gradient map, and compensation calculation is performed based on the phase error to obtain the phase correction sub-map corresponding to the first two-dimensional imaging sub-map.

[0057] A phase-corrected image is generated based on multiple phase-corrected sub-images corresponding to fixed distances.

[0058] Optionally, in one possible implementation, the first processing module is further configured to:

[0059] The first two-dimensional imaging sub-image is divided into multiple overlapping blocks; wherein, a single overlapping region occupies half of the overlapping block;

[0060] The center of each overlapping block is moved to the pixel with the highest signal strength in the overlapping block, resulting in multiple shifted overlapping blocks.

[0061] Optionally, in one possible implementation, the first processing module is further configured to:

[0062] Based on each shifted overlapping block, the center row and center column of the shifted overlapping block are obtained, and windowing calculation is performed based on the center row and center column to obtain the windowed overlapping block;

[0063] Phase gradient calculation is performed on each windowed overlapping block to obtain the phase gradient value of each windowed overlapping block;

[0064] The phase gradient values ​​of each windowed overlapping block are interpolated and mapped to the position of the windowed overlapping block in the first two-dimensional imaging sub-image to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image.

[0065] Optionally, in one possible implementation, the second processing module is further configured to:

[0066] For each fixed distance direction of the phase-corrected image, the second two-dimensional imaging sub-image corresponding to the phase-corrected image is extracted from the phase-corrected image;

[0067] The second two-dimensional imaging sub-image is uniformly divided into multiple basic sub-block images of the same size;

[0068] For each basic sub-block image, the spectral width and energy values ​​of the basic sub-block image are calculated; where the spectral width represents the signal frequency dispersion of the basic sub-block image; and the energy value represents the target reflection intensity of the basic sub-block image, where the target reflection intensity refers to the core target of MIMO radar imaging.

[0069] Based on the spectral width and energy values, each basic sub-block image is adaptively refined to obtain multiple sub-block images.

[0070] Optionally, in one possible implementation, the second processing module is further configured to:

[0071] Set a sliding window of a preset size in each basic sub-block image;

[0072] Perform a two-dimensional Fourier transform on each sliding window to obtain the wavenumber domain signal corresponding to the sliding window;

[0073] The spectral width and energy values ​​of each basic sub-block image are calculated based on the wavenumber domain signal.

[0074] Optionally, in one possible implementation, the second processing module is further configured to:

[0075] Based on the spectral width and energy values ​​of multiple basic sub-block images, the energy threshold and spectral width threshold of the second two-dimensional imaging sub-image are calculated.

[0076] For each basic sub-block image, when the spectral width value is greater than the spectral width threshold or the energy value is greater than the energy threshold, the basic sub-block image is divided into multiple first-level refined sub-blocks;

[0077] For each first-level refinement sub-block, if the spectral width value of the first-level refinement sub-block is greater than the spectral width threshold, or the energy value of the first-level refinement sub-block is greater than the energy threshold, the first-level refinement sub-block is divided into multiple second-level refinement sub-blocks.

[0078] Multiple sub-block images are obtained based on multiple basic sub-block images, multiple first-level thinning sub-blocks, and multiple second-level thinning sub-blocks.

[0079] Optionally, in one possible implementation, the fourth processing module is further configured to:

[0080] Each windowed sub-block image is expanded outwards to obtain an expanded sub-block image; there are overlapping areas between the expanded sub-block images.

[0081] For every two adjacent dilated sub-block images, determine two target window functions corresponding to the two dilated sub-block images;

[0082] Based on the preset fusion strategy and the function types corresponding to the two target window functions, the target fusion method of two adjacent expanded sub-block images is determined.

[0083] Based on the fusion method corresponding to each adjacent expanded sub-block image, fusion is performed on each overlapping region to obtain a fused image.

[0084] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0085] The memory stores instructions that the computer executes;

[0086] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect above and various possible implementations of the first aspect.

[0087] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and various possible implementations thereof.

[0088] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and various possible implementations thereof.

[0089] This application provides a near-field MIMO radar sidelobe suppression method based on phase gradient autofocus and spatial adaptive time-frequency window function. The method acquires the MIMO radar echo signal of the target scene and generates a wavenumber domain image corresponding to the radar echo signal. Phase gradient autofocus processing is performed on the wavenumber domain image to obtain a phase-corrected image with phase deviation eliminated. Based on the phase-corrected image, spatial adaptive partitioning is performed to obtain multiple sub-block images; window function matching is performed on the sub-block images to determine the target window function corresponding to each sub-block image. Windowing calculation is performed on each sub-block image and its corresponding target window function to obtain a windowed sub-block image; this windowed sub-block image is an image that has already undergone sidelobe suppression. Multiple windowed sub-block images are fused according to a preset fusion strategy to obtain a fused image; secondary phase gradient focusing is performed on the fused image to obtain the target imaging result of the near-field MIMO radar. Compared with existing technologies, this application first uses phase gradient autofocus to eliminate phase deviation in beam images and improve the clarity of image boundaries; then, it performs window function matching and windowing on the adaptively divided sub-block images to obtain images with sidelobe suppression; and then uses image fusion and secondary phase gradient autofocus for further image refinement, thereby achieving the technical effect of improving the precision of imaging results. Attached Figure Description

[0090] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0091] Figure 1 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 1 ;

[0092] Figure 2 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 2 ;

[0093] Figure 3 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 3 ;

[0094] Figure 4 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 4 ;

[0095] Figure 5 A schematic diagram of the near-field MIMO radar sidelobe suppression device based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application;

[0096] Figure 6 A schematic diagram of the structure of the electronic device provided in this application.

[0097] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0098] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0099] First, let's explain the terms used in this application:

[0100] Multiple-Input Multiple-Output (MIMO) refers to a radar architecture that uses multiple transmitting and receiving antennas to operate simultaneously.

[0101] Fourier Transform (FT): In this application, it refers to a mathematical tool used for signal domain transformation, which converts the spatial domain signal of radar echo into the wavenumber domain signal.

[0102] Two-Dimensional Discrete Fourier Transform (2D-DFT): refers to the two-dimensional extension of the Fourier transform, which is used in this application to process two-dimensional imaging sub-images or sliding window signals.

[0103] Three-Dimensional Inverse Discrete Fourier Transform (3D-IDFT): refers to the mathematical operation opposite to the Fourier Transform, which in this application is used to convert the processed wavenumber domain signal into an imaging result in the spatial domain.

[0104] The existing technology for sidelobe suppression of near-field MIMO radar involves using a preset window function to perform windowing calculations on the wavenumber domain image after the radar echo signal is converted, thereby achieving sidelobe suppression.

[0105] However, existing radar sidelobe suppression methods use fixed window functions; but the wavenumber domain signals in existing technologies are not uniformly distributed, resulting in false targets or main lobe blurring during sidelobe suppression; thus, existing technologies suffer from low imaging clarity.

[0106] To address the aforementioned technical problems, this application proposes the following technical concept: The acquired MIMO radar echo signal of the target scene is converted into a wavenumber domain image; phase gradient autofocus processing is performed on the wavenumber domain image to eliminate phase deviation and improve image boundary sharpness, resulting in a phase-corrected image. The phase-corrected image is spatially adaptively divided into multiple sub-block images; window function matching is performed on each sub-block image to determine its corresponding target window function, and the target window function is used to window the sub-block image, resulting in a windowed sub-block image, thereby achieving sidelobe suppression. Multiple windowed sub-block images are fused to obtain a fused image, and a secondary phase gradient autofocus processing is performed on the fused image to obtain the target imaging result. Compared with existing technologies, this application improves the accuracy of sidelobe suppression through window function matching and enhances the sharpness of imaging boundaries through phase gradient autofocus, thus achieving the technical effect of improving the clarity of the imaging result.

[0107] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0108] Figure 1 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 1 ,like Figure 1 As shown, the method includes:

[0109] S101. Acquire the MIMO radar echo signal of the target scene and generate a wavenumber domain image based on the MIMO radar echo signal.

[0110] In this step, MIMO radar refers to radar that utilizes spatial diversity to improve target detection performance by simultaneously operating multiple transmitting and receiving antennas. The MIMO radar echo signal refers to the electromagnetic signal carrying target information, which is captured by the receiving antenna after the electromagnetic wave emitted by the transmitting antenna is reflected from the target. Target information includes, but is not limited to, position information, shape information, and scattering coefficients. The wavenumber domain is isomorphic to the spatial frequency domain and reflects the oscillation frequency of electromagnetic waves in space. The wavenumber domain image of the MIMO radar echo signal refers to obtaining a three-dimensional image or two-dimensional slice with wavenumber as the coordinate axis by performing a spatial Fourier transform on the echo signal. Each point in the wavenumber domain image corresponds to the spatial frequency components of the target in the x and y directions, and its amplitude represents the intensity of that frequency component. In near-field scenarios, the relationship between the beam and the target position needs to be accurately mapped from the wavenumber domain to the spatial domain through spherical wave correction. Spherical wave correction can be achieved through range compensation or phase weighting.

[0111] Alternatively, one possible implementation for generating wavenumber domain images is as follows:

[0112] a1. Generating wavenumber domain signals based on MIMO radar echo signals;

[0113] Assuming the radar element located at (x,y) receives a signal with wavenumber k, s(x,y,k), then a Fourier transform is first performed along the x and y directions of the radar array to obtain the wavenumber domain signal, as shown in Equation 1:

[0114]

[0115] Where s(x,y,k) refers to the original radar echo signal; x and y refer to the radar element coordinates; k refers to the echo wavenumber; FT 2DThis refers to the Fourier transform. S(k) x ,k y (k) refers to the wavenumber domain signal; k x ,k y This refers to the wavenumber coordinates along the x and y directions.

[0116] a2. Compensate for the distance between the target and the radar and eliminate distance migration.

[0117] Assuming the distance between the radar array and the imaging target surface is Z, the wavenumber domain signal of the imaging target surface can be obtained by performing range compensation on the received wavenumber domain signal, as shown in Formula 2:

[0118]

[0119] Among them, F(k) x ,k y (k) refers to the wavenumber domain signal after distance compensation; k z It refers to the wavenumber coordinate along the z-direction; Z refers to the distance between the radar array and the imaging target surface.

[0120] And k z satisfy The F(k) obtained from this x ,k y ,k) in k z The non-uniform distribution along the axis physically corresponds to the range migration caused by the different distances between each radar element and the target. This range migration needs to be eliminated through Stolt interpolation to obtain k. z Axially uniform F(k) x ,k y ,k).

[0121] a3. The wavenumber domain image after coarse imaging is obtained by processing the wavenumber domain signal through inverse Fourier transform.

[0122] After distance compensation and distance migration correction, the wavenumber domain signal is then subjected to a three-dimensional inverse Fourier transform to obtain the scattering function distribution f(x,y,z) in the imaging space, as shown in Equation 3:

[0123]

[0124] Where f(x,y,z) refers to the scattering function distribution of the imaging target in space; This refers to the three-dimensional inverse Fourier transform; F(k) x ,k y ,k zThis refers to the wavenumber domain signal after distance compensation. The resulting wavenumber domain image is a coarse image, which requires further compensation for phase errors and post-processing to address main lobe expansion and side lobe leakage issues.

[0125] S102. Perform phase gradient autofocusing processing based on the wavenumber domain image to obtain a phase-corrected image.

[0126] In this step, phase gradient autofocus is used to eliminate phase deviations in wavenumber domain images and improve the clarity of image boundaries. Phase gradient autofocus works by utilizing the phase characteristics of strong scattering points in radar echo signals, iteratively estimating the phase error gradient, reconstructing the complete phase error, and compensating for it, thereby eliminating phase deviations.

[0127] Optionally, the phase gradient self-focusing can be achieved as follows:

[0128] b1. Divide the coarse wavenumber domain image into two-dimensional sub-images according to a fixed distance and use overlapping sliding blocks.

[0129] b2. Perform cyclic displacement on each block, extract the data of the center row and center column, multiply it with the window function, and calculate the phase gradient.

[0130] b3. Generate a global phase gradient map by interpolation and solve for the phase error using Fourier transform.

[0131] b4. Compensate for the phase error to obtain a phase-corrected image.

[0132] It should be noted that the specific implementation method of phase gradient self-focusing in this step is as follows: Figure 2 Further explanation will be provided in the embodiments shown, and will not be repeated here.

[0133] S103. Based on the phase-corrected image, perform spatial adaptive partitioning to obtain multiple sub-block images.

[0134] In this step, spatial adaptive partitioning refers to the spatial domain of the phase-corrected image; spatial adaptive partitioning means partitioning based on the image's own features; densely packed and textured areas in the image will be divided into smaller sub-blocks to preserve details; flat backgrounds and simple features in the image will be divided into larger sub-blocks to reduce redundant calculations; the boundaries of the sub-blocks will automatically fit the target edges or abrupt feature changes to avoid forcibly classifying heterogeneous features into the same sub-block.

[0135] Alternatively, the spatial adaptive partitioning can be implemented as follows:

[0136] c1. Divide the phase-corrected image into multiple basic sub-blocks of a fixed size, and set sliding windows within the basic sub-blocks.

[0137] c2. The spectral width and energy values ​​of the basic sub-blocks are obtained by sliding window calculation.

[0138] c3. Based on the spectral width and energy values ​​of the basic sub-blocks, the basic sub-blocks are dynamically refined to obtain multiple sub-block images.

[0139] It should be noted that, regarding the spatial adaptive partitioning of phase-corrected images, in the following... Figure 3 Further explanation will be provided in the embodiments shown, and will not be repeated here.

[0140] S104. Perform window function matching for each sub-block image to determine the target window function corresponding to each sub-block image, and window the sub-block image based on the target window function to obtain the windowed sub-block image.

[0141] In this step, the window function is used to perform weighted calculations on the sub-block images, improving the clarity of the sub-block image boundaries.

[0142] For example, the types of window functions are: Hamming window, Blackman window, and Kaiser window. The expressions for each window function are shown in Equation 4-6:

[0143]

[0144]

[0145]

[0146] Among them, w hamm (n) refers to the Hamming window; w balck (n) refers to the Blackman window; w kaiser (n) refers to the Kaiser window; n refers to the one-dimensional coordinates. N refers to the window length, and I0(•) refers to the zeroth-order Bessel function of the first kind. Among the three types of window functions, the Kaiser window has a parameter β that needs to be set, while the Hamming and Blackman windows have no parameters. The appropriate window function and parameters are selected based on the statistical characteristics of the sub-blocks. The statistical characteristics of the sub-blocks refer to the spectral width and energy values ​​of the sub-block image.

[0147] The appropriate window function is selected based on the characteristics of the sub-block image. The window function matching method can be as follows:

[0148] d1. Small spectral width and low energy sub-blocks: Hamming window. The spectrum of this type of sub-block image is concentrated, but the overall scattering target is not strong. Choosing a Hamming window with a narrow main lobe can maintain spatial resolution to the greatest extent while ensuring sidelobe suppression.

[0149] d2. Small spectral width, high energy sub-blocks: Blackman window. The images of this type of sub-block have high scattering target intensity, and sidelobe leakage is more likely to obscure weak targets. The Blackman window with high sidelobe suppression can effectively avoid interference from sidelobe pairs of strong scattering points.

[0150] d3, wide spectral width, low energy sub-block: Kaiser window, parameters set to... ,in Here, k is the base parameter, and k is the sensitivity gain. This type of sub-block image uses a wide main lobe window, and the parameters are adaptively adjusted according to the spectral width. When the width is large, the main lobe width is sacrificed to obtain stronger side lobe suppression and prevent wide-spectrum leakage.

[0151] d4. Wide spectral width and high energy sub-blocks: Kaiser window, with parameters fixed at the maximum value. This type of sub-block image contains both high-frequency components and high-energy targets, which easily leads to severe sidelobe leakage. A high-parameter Kaiser window can effectively suppress sidelobes.

[0152] In this step, after selecting the window function, a two-dimensional window is constructed. The windowed signal is obtained by multiplying the signal with the wavenumber domain signal corresponding to the sub-block image. This refers to a windowed sub-block image. Here, (i,j) represents the spatial coordinates of the sub-block image, where i corresponds to the x-direction and j corresponds to the y-direction. x (i) refers to the one-dimensional window function in the x-direction, w y (j) refers to the one-dimensional window function in the y-direction. W(i,j) refers to the two-dimensional window function, which is constructed by multiplying the one-dimensional window functions in the x and y directions.

[0153] S105. Based on a preset fusion strategy, multiple windowed block images are fused to obtain a fused image.

[0154] Alternatively, one possible way to obtain the fused image is as follows:

[0155] S1051. Expand each windowed sub-block image outward to obtain an expanded sub-block image.

[0156] In this step, there are overlapping regions between the various expanded sub-block images. These overlapping regions are due to artifacts that occur at the boundaries of the sub-blocks when different windowed sub-block images use different target window functions, caused by rapid changes in windowing weights. To suppress these artifacts, the windowed sub-block images are expanded outwards by a fixed number of rows and columns to form expanded sub-block images, resulting in overlapping regions between these expanded sub-block images.

[0157] For example, a windowed sub-block image has a size of 32×32 pixels, which is expanded outward by 8 pixels to obtain an expanded sub-block image of 48×48 pixels. The windowed sub-block image to its right is also expanded by 8 pixels, and the two expanded sub-block images form a 16-pixel overlap area at the boundary.

[0158] S1052. For every two adjacent expanded sub-block images, determine the two target window functions corresponding to the two expanded sub-block images.

[0159] In this step, the target window function refers to the target window function used for the windowed sub-block image corresponding to the expanded sub-block image. The characteristics of the target window function directly affect the local features of the expanded sub-block image, such as resolution and noise level. The purpose of determining the target window function is to provide a basis for subsequent image fusion. The fusion strategy for the overlapping areas of sub-block images processed by different window functions needs to match the characteristics of the window function.

[0160] For example, in two adjacent expanded sub-block images, the expanded sub-block image on the left corresponds to a scene with a small spectral width and high energy; the expanded sub-block image on the right corresponds to a scene with a large spectral width and low energy; by querying the processing records of the sub-blocks, the target window functions for the two are obtained as Blackman window and Kaiser window, respectively.

[0161] S1053. Based on the preset fusion strategy and the function types corresponding to the two target window functions, determine the target fusion method of the two adjacent expanded sub-block images.

[0162] In this step, the preset fusion strategies include: for two sub-blocks with inconsistent window changes, simple weighted fusion is used; for two sub-blocks with drastic window changes, Poisson fusion is used; for pixels at the four corners of a sub-block that may be simultaneously located in overlapping areas covered by three or four sub-blocks, two-dimensional weighted fusion or two-dimensional Poisson fusion is used to achieve sub-block fusion.

[0163] S1054. Based on the fusion method corresponding to each adjacent expanded sub-block image, fuse each overlapping region to obtain a fused image.

[0164] In this step, by selectively fusing overlapping areas, the boundary differences between sub-blocks are eliminated, ensuring that the target is continuous and smooth in the global image, while preserving the local processing advantages of each sub-block, and finally generating a complete image without stitching marks.

[0165] For example, the method for obtaining a fused image can be as follows:

[0166] e1. For two sub-blocks with inconsistent window changes: between the Hamming window and the Blackman window; and between two Kaiser windows with a β value difference of no more than 2, a simple weighted fusion is used, along k... xTaking the fusion of directions as an example, the fusion method is shown in Formula 7:

[0167]

[0168] in, It refers to the wavenumber domain signal of the overlapping region after fusion, that is, the representation of the expanded sub-block image after fusion in the wavenumber domain; These refer to the spatial frequencies in the x and y directions, respectively, and are used to reflect the oscillation characteristics of the signal in space.

[0169] This refers to two adjacent expanded sub-block images; These refer to the weighting parameters, which correspond to the contribution ratios of sub-blocks p and q in the overlapping region, respectively. The calculation method for the weighting parameters is shown in Formula 8:

[0170]

[0171] in, X refers to the minimum coordinate of the overlapping area of ​​the two sub-windows along the x-direction; X refers to the width of the overlapping area along the x-direction. This refers to the distance from a pixel x within the overlapping region to its actual boundary. The distance is used to reflect the relative position of the pixel in the overlapping area.

[0172] e2. For two sub-blocks with drastic window changes: the boundary between the Hamming window, the Blackman window, and the Kaiser window; two Kaiser windows with a β value difference exceeding 2; Poisson fusion is used, and the solution is shown in Equation 9:

[0173]

[0174] in, This refers to the overlapping area, that is, the area that needs to be merged; It refers to the boundary of the overlapping area, which is the dividing line between the merged area and the non-overlapping area. This refers to the wavenumber domain image within the overlapping region after sub-block fusion; This refers to the reference gradient field in The directional components are shown in Formula 10:

[0175]

[0176] in, This refers to the wavenumber domain image within the overlapping region that has not been fused into sub-blocks; This refers to the component of the reference gradient field in the x-direction, representing the signal difference between the position (i,j) and the right-side adjacent pixel (i+1,j). This refers to the component of the reference gradient field in the y-direction, representing the signal difference between position (i,j) and its adjacent pixel (i,j+1) below.

[0177] e3. For pixels at the four corners of the expanded sub-block image that may be simultaneously located in the overlapping area covered by three or four sub-blocks, two-dimensional weighted fusion or two-dimensional Poisson fusion is used to achieve sub-block fusion.

[0178] e5. The results of fusing sub-blocks from different two-dimensional planes are stitched together to obtain the final three-dimensional image data, which is the complete wavenumber domain image after sub-block fusion. The complete wavenumber domain image refers to the fused image calculated in this step.

[0179] S106. Perform secondary phase gradient autofocusing processing on the fused image to obtain the target imaging result.

[0180] In this step, secondary phase gradient autofocus refers to applying the phase gradient autofocus algorithm again after the fused image is generated. The purpose of this step is to eliminate phase errors caused by operations such as sub-block expansion and overlapping region weighting during the fusion process.

[0181] Optionally, the process of secondary phase gradient self-focusing is as follows:

[0182] f1. Perform an inverse Fourier transform on the fused image to obtain a refined imaging image; as shown in Equation 11:

[0183]

[0184] in, This refers to the final imaging result image, i.e., the fine imaging image; The final imaging result in the wavenumber domain, i.e., the fused image. This refers to the three-dimensional inverse Fourier transform; It refers to the wave number in the three directions of x, y, and z.

[0185] f2. Perform phase gradient autofocus on the finely imaged image to obtain the final imaging result.

[0186] This application provides a near-field MIMO radar sidelobe suppression method based on phase gradient autofocus and spatial adaptive time-frequency window function. This method acquires the MIMO radar echo signal of the target scene and generates a wavenumber domain image corresponding to the radar echo signal. Phase gradient autofocus processing is performed on the wavenumber domain image to obtain a phase-corrected image with phase deviation eliminated. Based on the phase-corrected image, spatial adaptive partitioning is performed to obtain multiple sub-block images. Window function matching is performed on the sub-block images to determine the target window function corresponding to each sub-block image. Windowing calculation is performed on each sub-block image and its corresponding target window function to obtain a windowed sub-block image; this windowed sub-block image is an image that has already undergone sidelobe suppression. Multiple windowed sub-block images are fused according to a preset fusion strategy to obtain a fused image. Secondary phase gradient focusing is performed on the fused image to obtain the target imaging result of the near-field MIMO radar. Compared with existing technologies, this application first uses phase gradient autofocus to eliminate phase deviation in beam images and improve the clarity of image boundaries; then, it performs window function matching and windowing on the adaptively divided sub-block images to obtain images with sidelobe suppression; and then uses image fusion and secondary phase gradient autofocus for further image refinement, thereby achieving the technical effect of improving the precision of imaging results.

[0187] Figure 2 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 2 Based on the above embodiments, this embodiment further... Figure 1 The phase gradient self-focusing in step S102 of the illustrated embodiment will be further explained as follows: Figure 2 As shown, the method includes:

[0188] S201. For each fixed distance direction in the wavenumber domain image, extract the first two-dimensional imaging sub-image corresponding to the fixed distance direction from the wavenumber domain image.

[0189] In this step, the range direction corresponds to the radial range dimension between the radar and the target. A fixed range direction means selecting a specific distance and slicing the three-dimensional wavenumber domain image along that range. The first two-dimensional imaging sub-image refers to the two-dimensional image obtained after slicing, which only contains the target's spatial frequency information in that fixed range direction.

[0190] For example, a three-dimensional wavenumber domain image has a size of 512 (range) × 256 (azimuth) × 256 (elevation), meaning it contains 512 range axes. Selecting a range axis index r = 128 and slicing along this dimension yields a 256 × 256 first two-dimensional imaging sub-image. This sub-image only reflects the spatial frequency distribution of the target's azimuth and elevation axes at the range index r = 128.

[0191] S202. Based on the first two-dimensional imaging sub-image, perform block processing to obtain multiple overlapping blocks.

[0192] In this step, block processing refers to dividing the first two-dimensional imaging sub-image into several local regions, each block containing local target information. Overlapping block processing refers to blocks where adjacent blocks have overlapping areas, avoiding the block boundaries from interrupting the continuity of the target.

[0193] Alternatively, one possible implementation of obtaining multiple overlapping blocks is as follows:

[0194] S2021. Divide the first two-dimensional imaging sub-image into multiple overlapping blocks.

[0195] In this step, a single overlapping region occupies half of the overlapping block.

[0196] For example, taking a 256×256 first two-dimensional imaging sub-image as an example; setting the block size to 32×32 pixels and the overlapping area to 16×16 pixels, sliding the blocks along the azimuth and height, a total of 225 overlapping blocks can be obtained.

[0197] S2022. Move the center of each overlapping block to the pixel with the highest signal strength in the overlapping block to obtain multiple shifted overlapping blocks.

[0198] In this step, moving the overlapping block refers to transforming the pixel with the highest signal strength in the overlapping block to the center position of the overlapping block through cyclic shifting.

[0199] For example, for each 32×32 overlapping block, the signal strength of the internal pixels is calculated, and the pixel coordinates corresponding to the maximum intensity are found by iterative calculation. The block center is then moved from its original position to the position corresponding to the pixel coordinates to obtain the shifted overlapping block.

[0200] It should be noted that the purpose of this step is to ensure that the center of the block is aligned with the strong scattering point, thereby improving the reliability of subsequent phase gradient estimation.

[0201] S203. Calculate the phase gradient based on multiple overlapping blocks to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image.

[0202] In this step, the phase gradient refers to the rate of change of phase with spatial location, used to reflect the trend of local phase change. The global gradient map refers to the phase gradient distribution covering the entire region of the first two-dimensional imaging sub-map after integrating the local phase gradients of all overlapping blocks.

[0203] Alternatively, one possible implementation of obtaining the global gradient map is as follows:

[0204] S2031. Based on each shifted overlapping block, obtain the center row and center column of the shifted overlapping block, and perform windowing calculation based on the center row and center column to obtain the windowed overlapping block.

[0205] In this step, windowed overlapping blocks refer to the windowed overlapping blocks obtained by windowing calculation using a pre-defined window function and the center row and center column; windowed overlapping blocks refer to the overlapping blocks after edge noise has been suppressed.

[0206] For example, the way to obtain windowed overlapping blocks can be:

[0207] For each fixed distance Extract the first two-dimensional imaging sub-image Divide the subgraph evenly into N overlapping blocks. The blocks are slid in an overlapping manner, with an overlap step size of half the block length. For each overlapping block, the pixel with the highest signal strength is cyclically shifted to the center position of the block, and the center row and center column data are extracted. After multiplying each of these by a one-dimensional Hamming window, the windowed overlapping blocks are obtained, as shown in Formula 12:

[0208]

[0209] in This refers to the one-dimensional data of the center row extracted after the nth overlapping block is cyclically shifted; This refers to the one-dimensional data of the center column extracted after the nth overlapping block is cyclically shifted. w(x) refers to the one-dimensional Hamming window that matches the center row data; w(y) refers to the one-dimensional Hamming window that matches the center column data. This refers to the windowed center row data obtained by multiplying the center row data by the Hamming window. This refers to the windowed center column data resulting from multiplying the center column by the Hamming window. The windowed center row data and the windowed center column data together form the windowed overlapping block.

[0210] S2032. Calculate the phase gradient based on each windowed overlapping block to obtain the phase gradient value of each windowed overlapping block.

[0211] In this step, calculating the phase gradient of the windowed overlapping blocks refers to estimating the phase gradient using the central difference method. The phase gradient value is calculated as shown in Equation 13:

[0212]

[0213] in, This refers to the phase gradient value of the nth windowed overlapping block in the horizontal direction, i.e., the phase gradient value of the center row; This refers to the phase gradient value in the vertical direction of the nth windowed overlapping block, i.e., the phase gradient value of the center column. Im{·} refers to the imaginary part operator for complex numbers; ln(·) refers to the natural pair operator for complex numbers. It refers to the windowing signal of the pixel to the right of index x in the center row of the nth windowed overlapping block; It refers to the windowing signal of the pixel to the right and left of index x in the center row of the nth windowed overlapping block. It refers to the windowing signal of the pixel adjacent to the y-index in the center column of the nth windowed overlapping block; It refers to the windowing signal of the pixel above the y-index in the center column of the nth windowed overlapping block.

[0214] S2033. The phase gradient values ​​of each windowed overlapping block are interpolated and mapped to the position of the windowed overlapping block in the first two-dimensional imaging sub-image to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image.

[0215] In this step, the global gradient map of the first two-dimensional imaging sub-image is calculated as follows: the phase gradient values ​​of the two directions corresponding to the windowed overlapping blocks are interpolated back to their positions in the entire image, generating gradient maps in the x and y directions. and The gradient values ​​in both directions are determined by interpolation within the corresponding windowed overlapping blocks, and are 0 outside the blocks. At each pixel location, the average phase gradient from all blocks is calculated, as shown in Equation 14.

[0216]

[0217] in, It refers to the global gradient map of the first two-dimensional imaging sub-image in the x-direction. It is a two-dimensional matrix in the spatial domain, used to reflect the average rate of change of the phase of each pixel in the entire image in the x-direction. It refers to the global gradient map in the y-direction of the first two-dimensional imaging sub-image, which reflects the average rate of change of the phase of each pixel in the entire image in the y-direction. This refers to the local gradient map of the nth windowed overlapping block in the x-direction, which is generated by interpolating and mapping the phase gradient value of the nth windowed overlapping block in the x-direction to the corresponding position of the first two-dimensional imaging sub-map. This refers to the local gradient map in the y-direction of the nth windowed overlapping block, generated by interpolating and mapping the phase gradient value in the y-direction of the nth windowed overlapping block to the corresponding position in the first two-dimensional imaging sub-image. N(x,y) refers to the number of windowed overlapping blocks that simultaneously cover pixel (x,y) in the first two-dimensional imaging sub-image. This is mainly because the blocks are overlapping sliding blocks, and a single pixel may be covered by multiple blocks.

[0218] S204. The phase error is calculated based on the global gradient map, and compensation calculation is performed based on the phase error to obtain the phase correction sub-map corresponding to the first two-dimensional imaging sub-map.

[0219] In this step, phase error refers to the global phase error corresponding to the global gradient map; that is, the deviation between the actual signal phase and the ideal focusing phase. The global gradient map is essentially the partial derivative of the global phase error in the x and y directions; the relationship between the phase error and the partial derivative of the global gradient is shown in Equation 15:

[0220]

[0221] in, This refers to the global phase error of the first two-dimensional imaging sub-image; It refers to the first-order partial derivative of the global phase error with respect to the x-direction, i.e., the global gradient map in the x-direction; This refers to the first-order partial derivative of the global phase error with respect to the y-direction, which is the global gradient map in the y-direction. The remaining parameters are explained in Equation 14.

[0222] In this step, the phase error can be solved using the Fourier transform method. Specifically, the gradient field divergence is calculated using Equation 16:

[0223]

[0224] The Fourier transform of the divergence is obtained as follows The phase error can be solved in the frequency domain using formula 17:

[0225]

[0226] Based on the phase error, an inverse Fourier transform is performed to obtain... .

[0227] in, It refers to the two-dimensional divergence of the global gradient field G, which describes the degree of convergence or divergence of the gradient field at a certain point. This refers to the frequency domain representation of the gradient field divergence after a two-dimensional Fourier transform, used to reflect the distribution of divergence in the frequency domain. This refers to the representation of global phase error in the frequency domain; It is the equivalent term of the Laplace operator in the frequency domain. This refers to the spatial domain global phase error obtained through the inverse Fourier transform. This refers to the first-order partial derivative of the global gradient map with respect to the x-direction; It refers to the first-order partial derivative of the global gradient map with respect to the y-direction.

[0228] In this step, the phase correction sub-graph is calculated using the phase error compensation method, as shown in Formula 18:

[0229]

[0230] in, This refers to the image before phase gradient autofocusing, i.e., the first two-dimensional imaging sub-image; This refers to the phase correction sub-graph after phase gradient self-focusing processing. This refers to the compensating phase factor used to offset phase errors; This refers to the global phase error.

[0231] S205. Generate a phase-corrected image based on multiple phase-corrected sub-images corresponding to fixed distance directions.

[0232] In this step, the multiple phase-corrected sub-images corresponding to fixed range directions refer to the process of performing phase gradient autofocus on all range directions in the wavenumber domain image to obtain multiple two-dimensional phase-corrected sub-images; and stitching all the correction sub-images together in the order of range directions to restore the complete three-dimensional image, which is the phase-corrected image.

[0233] Figure 3 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 3 This embodiment is for Figure 1 The spatial adaptive partitioning of step S103 in the illustrated embodiment will be further explained as follows: Figure 3 As shown, the method includes:

[0234] S301. For each fixed distance direction of the phase-corrected image, extract the second two-dimensional imaging sub-image corresponding to the phase-corrected image from the phase-corrected image.

[0235] In this step, the phase-corrected image is a global image that has undergone one phase gradient autofocusing process, eliminating most of the phase distortion. It contains target information in all range, azimuth, and altitude directions. The logic for the fixed range direction is the same as described above. Figure 2 The logic of step S201 in the embodiment is consistent, which means fixing the z-coordinate of the three-dimensional image and retaining only the two-dimensional data of the x and y planes. The second two-dimensional imaging sub-image refers to the planar two-dimensional image extracted after fixing the z value. Its size and x and y dimensions are consistent with the phase-corrected image. It is used to reflect the orientation and height distribution of the target at that distance and has eliminated the previous phase error.

[0236] S302. Divide the second two-dimensional imaging sub-image into multiple basic sub-block images of the same size.

[0237] In this step, uniform division refers to dividing the second two-dimensional imaging sub-image into several local regions of identical size according to a fixed dimension. During the division process, there is no overlap or only minimal overlap, ensuring that the basic sub-block image is completely covered without repetition. The basic sub-block image refers to the largest and smallest initial local block after uniform division, which is the basic unit for subsequent calculations of spectral width, energy, and refined block segmentation.

[0238] S303. For each basic sub-block image, calculate the spectral width and energy value of the basic sub-block image.

[0239] In this step, the spectral width value represents the signal frequency dispersion of the basic sub-block image; the energy value represents the target reflection intensity of the basic sub-block image, and the target reflection intensity refers to the core target of MIMO radar imaging.

[0240] Alternatively, one possible way to calculate the energy value and spectral width value is as follows:

[0241] S3031. Set a sliding window of a preset size in each basic sub-block image.

[0242] In this step, setting a preset-sized sliding window refers to setting a small-sized moving window in the basic sub-block image. By sliding and calculating the features of multiple windows, the average value is taken as the overall feature of the sub-block, thus avoiding interference from local noise within the sub-block on feature calculation.

[0243] For example, the size of the basic sub-block image is 32×32 pixels; the size of the sliding window is set to 16×16 pixels, the sliding step is 8 pixels, and a total of 9 sliding windows can be obtained within each basic sub-block image.

[0244] S3032. Perform a two-dimensional Fourier transform on each sliding window to obtain the wavenumber domain signal corresponding to the sliding window.

[0245] In this step, a two-dimensional Fourier transform is performed on the spatial domain signal of each sliding window to obtain the corresponding wavenumber domain signal.

[0246] For example, the process of extracting the second two-dimensional imaging sub-image and calculating the wavenumber domain signal can be as follows: First, fix the distance to... Extract the second two-dimensional imaging sub-image The image is divided into L×L basic sub-blocks of a fixed size, where L refers to the size of the basic sub-block image. Within each basic sub-block image, a sliding window of size l×l is set, with l = L / 4 and a step size of l / 2. For each sliding window... The two-dimensional Fourier transform is calculated to the wavenumber domain to obtain the corresponding wavenumber domain signal. The specific calculation method is shown in Equation 19:

[0247]

[0248] in, : refers to the second two-dimensional imaging sub-image; It refers to the two-dimensional Fourier transform. : The set of pixels within the i-th and j-th sliding windows; This refers to the wavenumber domain signal corresponding to the i-th or j-th sliding window, which is used to reflect the spatial frequency distribution of the radar echo signal within that sliding window. These correspond to the spatial frequencies in the x and y directions, respectively.

[0249] S3033. The spectral width and energy values ​​of each basic sub-block image are calculated based on the wavenumber domain signal.

[0250] In this step, the energy value is calculated as follows: based on the wavenumber domain signal, the energy value of each sliding window in the basic sub-block image is calculated; the average energy value of multiple sliding windows in the basic sub-block image is then calculated to obtain the energy value of the basic sub-block image. Similarly, the spectral width value of the basic sub-block image is calculated. The method for calculating the energy value corresponding to the wavenumber domain signal of the sliding window is shown in Formula 20:

[0251]

[0252] in, This refers to the total energy in the wavenumber domain corresponding to the i-th and j-th sliding windows; for the explanation of the other parameters, please refer to Formula 19.

[0253] The spectral width of the sliding window is calculated as shown in Formula 21:

[0254]

[0255] in, This refers to the wavenumber domain spectral width of the i-th, j-th sliding window; Here, represents the centroid wavenumber; the other parameters are explained in Equations 19 and 20. The centroid wavenumber is calculated as shown in Equation 22.

[0256]

[0257] in, The centroid wavenumber refers to the energy centroid of the wavenumber domain signal. A small centroid wavenumber indicates that the signal is mainly concentrated in the low-frequency region, while a large wavenumber indicates that the signal contains more high-frequency components. For explanations of the other parameters, refer to formulas 19 to 21 above.

[0258] S304. Based on the spectral width and energy values, each basic sub-block image is adaptively refined to obtain multiple sub-block images.

[0259] In this step, adaptive refinement refers to determining whether the sub-blocks need to be further segmented based on the comparison results of the spectral width, energy value and preset threshold of the basic sub-block image.

[0260] Alternatively, one possible implementation of adaptive thinning to obtain multiple sub-block images is as follows:

[0261] S3041. Based on the spectral width and energy values ​​of multiple basic sub-block images, calculate the energy threshold and spectral width threshold of the second two-dimensional imaging sub-image.

[0262] In this step, the calculation of the energy threshold and spectral width threshold refers to calculating the average energy and average spectral width of multiple basic sub-block images, and then performing a weighted calculation based on the average energy and average spectral width to obtain the energy threshold and spectral width threshold.

[0263] For example, the energy threshold and spectral width threshold are calculated in the manner shown in Equation 23:

[0264]

[0265] in, The spectral width threshold refers to the spectral width threshold. This refers to the energy threshold; This refers to the standard deviation of the spectral width of the basic sub-block image; This refers to the average spectral width of the basic sub-block image; This refers to the average energy of the basic sub-block image; This refers to the weighting parameters used in the weighted calculation, which can be selected based on the required image quality (e.g., taking...). ).

[0266] S3042. For each basic sub-block image, when the spectral width value is greater than the spectral width threshold or the energy value is greater than the energy threshold, the basic sub-block image is divided into multiple first-level refined sub-blocks.

[0267] In this step, if either the spectral width or the energy value of the basic sub-block image exceeds its corresponding threshold, the basic sub-block image can be divided along the x and y lines, with the division scale being half of the original scale of the basic sub-block image; thus, four first-level refined sub-blocks are obtained.

[0268] For example, if the size of the basic sub-block image is 32×32 pixels, then the size of the four first-level refined sub-blocks obtained by division is 16×16 pixels.

[0269] S3043. For each first-level refined sub-block, if the spectral width value of the first-level refined sub-block is greater than the spectral width threshold, or the energy value of the first-level refined sub-block is greater than the energy threshold, the first-level refined sub-block is divided into multiple second-level refined sub-blocks.

[0270] In this step, the spectral width and energy values ​​of the first-level refined sub-blocks are recalculated using the same sliding window wavenumber domain method. A threshold comparison is performed based on the calculated spectral width and energy values. If either exceeds the threshold, further refinement is performed to obtain the second-level refined sub-blocks. The refinement size used in the second-level refined sub-blocks is half that of the first-level refined sub-blocks, resulting in one first-level refined sub-block corresponding to four second-level refined sub-blocks.

[0271] It should be noted that the refinement limit is two times; once a second-level refined sub-block is obtained, no further refinement will be performed.

[0272] S3044. Based on multiple basic sub-block images, multiple first-level thinning sub-blocks, and multiple second-level thinning sub-blocks, multiple sub-block images are obtained.

[0273] In this step, the multiple sub-block images obtained here refer to the multiple sub-block images corresponding to multiple second two-dimensional imaging sub-images with fixed distance directions.

[0274] Figure 4 A flowchart illustrating the near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application. Figure 4 ,like Figure 4 As shown, the method includes:

[0275] A1. Convert the acquired near-field MIMO radar echo signal into a wavenumber domain signal.

[0276] A2. By processing the wavenumber domain signal through distance compensation and Stolt difference, the wavenumber domain signal after distance compensation and distance migration correction is obtained.

[0277] A3. The wavenumber domain image after coarse imaging is obtained by processing the wavenumber domain signal through inverse Fourier transform.

[0278] A4. Based on the wavenumber domain image, perform phase gradient autofocusing, preliminary segmentation, windowing, and phase correction processing to obtain a phase-corrected image.

[0279] A5. Based on energy and spectral width thresholds, the phase-corrected image is spatially adaptively divided to obtain multiple sub-block images.

[0280] A6. Perform window function matching and windowing on the sub-block image to obtain the windowed sub-block image.

[0281] A7. Based on weighted fusion and Poisson fusion, windowed block images are fused to obtain a fused image, and then proceed to step A3 for secondary phase gradient autofocus processing.

[0282] A8. Output the imaging results after secondary phase gradient autofocus processing.

[0283] Figure 5 A schematic diagram of the near-field MIMO radar sidelobe suppression device based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this application is shown below. Figure 5 As shown, the near-field MIMO radar sidelobe suppression device based on phase gradient self-focusing and spatial adaptive time-frequency window function provided in this embodiment includes:

[0284] The acquisition module 501 is used to acquire the MIMO radar echo signal of the target scene and generate a wavenumber domain image based on the MIMO radar echo signal;

[0285] The first processing module 502 is used to perform phase gradient autofocusing processing on the wavenumber domain image to obtain a phase-corrected image; the phase gradient autofocusing is used to eliminate phase deviation in the wavenumber domain image and improve the clarity of the image boundary.

[0286] The second processing module 503 is used to perform spatial adaptive partitioning based on the phase-corrected image to obtain multiple sub-block images;

[0287] The third processing module 504 is used to perform window function matching for each sub-block image, determine the target window function corresponding to each sub-block image, and window the sub-block image based on the target window function to obtain a windowed sub-block image; the window function is used to perform weighted calculation on the sub-block image to improve the clarity of the sub-block image boundary;

[0288] The fourth processing module 505 is used to fuse multiple windowed block images based on a preset fusion strategy to obtain a fused image;

[0289] The fifth processing module 506 is used to perform secondary phase gradient autofocus processing on the fused image to obtain the target imaging result.

[0290] Optionally, in one possible implementation, the first processing module 502 is further configured to:

[0291] For each fixed range direction in the wavenumber domain image, the first two-dimensional imaging sub-image corresponding to the fixed range direction is extracted from the wavenumber domain image;

[0292] Based on the first two-dimensional imaging sub-image, block processing is performed to obtain multiple overlapping blocks;

[0293] Phase gradient calculation is performed based on multiple overlapping blocks to obtain the global gradient map corresponding to the first two-dimensional imaging sub-map;

[0294] The phase error is calculated based on the global gradient map, and compensation calculation is performed based on the phase error to obtain the phase correction sub-map corresponding to the first two-dimensional imaging sub-map.

[0295] A phase-corrected image is generated based on multiple phase-corrected sub-images corresponding to fixed distances.

[0296] Optionally, in one possible implementation, the first processing module 502 is further configured to:

[0297] The first two-dimensional imaging sub-image is divided into multiple overlapping blocks; wherein, a single overlapping region occupies half of the overlapping block;

[0298] The center of each overlapping block is moved to the pixel with the highest signal strength in the overlapping block, resulting in multiple shifted overlapping blocks.

[0299] Optionally, in one possible implementation, the first processing module 502 is further configured to:

[0300] Based on each shifted overlapping block, the center row and center column of the shifted overlapping block are obtained, and windowing calculation is performed based on the center row and center column to obtain the windowed overlapping block;

[0301] Phase gradient calculation is performed on each windowed overlapping block to obtain the phase gradient value of each windowed overlapping block;

[0302] The phase gradient values ​​of each windowed overlapping block are interpolated and mapped to the position of the windowed overlapping block in the first two-dimensional imaging sub-image to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image.

[0303] Optionally, in one possible implementation, the second processing module 503 is further configured to:

[0304] For each fixed distance direction of the phase-corrected image, the second two-dimensional imaging sub-image corresponding to the phase-corrected image is extracted from the phase-corrected image;

[0305] The second two-dimensional imaging sub-image is uniformly divided into multiple basic sub-block images of the same size;

[0306] For each basic sub-block image, the spectral width and energy values ​​of the basic sub-block image are calculated; where the spectral width represents the signal frequency dispersion of the basic sub-block image; and the energy value represents the target reflection intensity of the basic sub-block image, where the target reflection intensity refers to the core target of MIMO radar imaging.

[0307] Based on the spectral width and energy values, each basic sub-block image is adaptively refined to obtain multiple sub-block images.

[0308] Optionally, in one possible implementation, the second processing module 503 is further configured to:

[0309] Set a sliding window of a preset size in each basic sub-block image;

[0310] Perform a two-dimensional Fourier transform on each sliding window to obtain the wavenumber domain signal corresponding to the sliding window;

[0311] The spectral width and energy values ​​of each basic sub-block image are calculated based on the wavenumber domain signal.

[0312] Optionally, in one possible implementation, the second processing module 503 is further configured to:

[0313] Based on the spectral width and energy values ​​of multiple basic sub-block images, the energy threshold and spectral width threshold of the second two-dimensional imaging sub-image are calculated.

[0314] For each basic sub-block image, when the spectral width value is greater than the spectral width threshold or the energy value is greater than the energy threshold, the basic sub-block image is divided into multiple first-level refined sub-blocks;

[0315] For each first-level refinement sub-block, if the spectral width value of the first-level refinement sub-block is greater than the spectral width threshold, or the energy value of the first-level refinement sub-block is greater than the energy threshold, the first-level refinement sub-block is divided into multiple second-level refinement sub-blocks.

[0316] Multiple sub-block images are obtained based on multiple basic sub-block images, multiple first-level thinning sub-blocks, and multiple second-level thinning sub-blocks.

[0317] Optionally, in one possible implementation, the fourth processing module 505 is further configured to:

[0318] Each windowed sub-block image is expanded outwards to obtain an expanded sub-block image; there are overlapping areas between the expanded sub-block images.

[0319] For every two adjacent dilated sub-block images, determine two target window functions corresponding to the two dilated sub-block images;

[0320] Based on the preset fusion strategy and the function types corresponding to the two target window functions, the target fusion method of two adjacent expanded sub-block images is determined.

[0321] Based on the fusion method corresponding to each adjacent expanded sub-block image, fusion is performed on each overlapping region to obtain a fused image.

[0322] The apparatus provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0323] Figure 6 A schematic diagram of the structure of the electronic device provided in this application. Figure 6As shown, the electronic device provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0324] In the specific implementation process, at least one processor 601 executes the computer execution instructions stored in the memory 602, causing at least one processor 601 to execute the above-mentioned near-field MIMO radar sidelobe suppression method or approach based on phase gradient self-focusing and spatial adaptive time-frequency window function.

[0325] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0326] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0327] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0328] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0329] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0330] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0331] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0332] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0333] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0334] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0335] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0336] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0337] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0338] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A near-field MIMO radar sidelobe suppression method based on phase gradient self-focusing and spatial adaptive time-frequency window function, characterized in that, include: Acquire the multi-input multi-output (MIMO) radar echo signal of the target scene, and generate a wavenumber domain image based on the MIMO radar echo signal; Phase gradient autofocusing is performed on the wavenumber domain image to obtain a phase-corrected image; the phase gradient autofocusing is used to eliminate phase deviation in the wavenumber domain image and improve the clarity of the image boundaries. Based on the phase-corrected image, spatial adaptive partitioning is performed to obtain multiple sub-block images; For each sub-block image, window function matching is performed to determine the target window function corresponding to each sub-block image, and the sub-block image is windowed based on the target window function to obtain a windowed sub-block image; The window function is used to perform weighted calculations on the sub-block image to improve the clarity of the sub-block image boundaries; Multiple windowed block images are fused based on a preset fusion strategy to obtain a fused image; The fused image is subjected to secondary phase gradient autofocus processing to obtain the target imaging result.

2. The method according to claim 1, characterized in that, The step of performing phase gradient autofocusing processing based on the wavenumber domain image to obtain a phase-corrected image includes: For each fixed distance direction in the wavenumber domain image, a first two-dimensional imaging sub-image corresponding to the fixed distance direction is extracted from the wavenumber domain image; Based on the first two-dimensional imaging sub-image, block processing is performed to obtain multiple overlapping blocks; Phase gradient calculation is performed based on multiple overlapping blocks to obtain the global gradient map corresponding to the first two-dimensional imaging sub-map; The phase error is calculated based on the global gradient map, and compensation calculation is performed based on the phase error to obtain the phase correction sub-map corresponding to the first two-dimensional imaging sub-map. The phase-corrected image is generated based on multiple phase-corrected sub-images corresponding to fixed distances.

3. The method according to claim 2, characterized in that, The block segmentation based on the first two-dimensional imaging sub-image yields multiple overlapping blocks, including: The first two-dimensional imaging sub-image is divided into multiple overlapping blocks; wherein, a single overlapping region in each overlapping block occupies half of the overlapping block; The center of each overlapping block is moved to the pixel with the highest signal strength in the overlapping block to obtain multiple shifted overlapping blocks.

4. The method according to claim 3, characterized in that, The step of calculating the phase gradient based on multiple overlapping blocks to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image includes: Based on each shifted overlapping block, the center row and center column of the shifted overlapping block are obtained, and windowing calculation is performed based on the center row and the center column to obtain windowed overlapping blocks; Phase gradient calculation is performed on each of the windowed overlapping blocks to obtain the phase gradient value of each of the windowed overlapping blocks; The phase gradient values ​​of each windowed overlapping block are interpolated and mapped to the position of the windowed overlapping block in the first two-dimensional imaging sub-image to obtain the global gradient map corresponding to the first two-dimensional imaging sub-image.

5. The method according to claim 1, characterized in that, The spatial adaptive partitioning based on the phase-corrected image yields multiple sub-block images, including: For each fixed distance direction of the phase-corrected image, a second two-dimensional imaging sub-image corresponding to the phase-corrected image is extracted from the phase-corrected image; The second two-dimensional imaging sub-image is uniformly divided into multiple basic sub-block images of the same size; For each of the basic sub-block images, the spectral width and energy values ​​of the basic sub-block image are calculated; wherein, the spectral width value represents the signal frequency dispersion of the basic sub-block image; and the energy value represents the target reflection intensity of the basic sub-block image, wherein the target reflection intensity refers to the core target of MIMO radar imaging. Based on the spectral width and energy values, each basic sub-block image is adaptively refined to obtain multiple sub-block images.

6. The method according to claim 5, characterized in that, The calculation of the spectral width and energy value of each basic sub-block image includes: A sliding window of a preset size is set in each of the basic sub-block images; Perform a two-dimensional Fourier transform on each of the sliding windows to obtain the wavenumber domain signal corresponding to the sliding window; The spectral width and energy values ​​of each basic sub-block are calculated based on the wavenumber domain signal.

7. The method according to claim 5, characterized in that, The basic sub-block image is adaptively refined based on the spectral width and energy values ​​to obtain multiple sub-block images, including: Based on the spectral width and energy values ​​of multiple basic sub-block images, the energy threshold and spectral width threshold of the second two-dimensional imaging sub-image are calculated. For each of the basic sub-block images, when the spectral width value is greater than the spectral width threshold, or the energy value is greater than the energy threshold, the basic sub-block image is divided into multiple first-level refined sub-blocks; For each of the first-level refined sub-blocks, when the spectral width value of the first-level refined sub-block is greater than the spectral width threshold, or when the energy value of the first-level refined sub-block is greater than the energy threshold, the first-level refined sub-block is divided into multiple second-level refined sub-blocks. Multiple sub-block images are obtained based on multiple basic sub-block images, multiple first-level thinning sub-blocks, and multiple second-level thinning sub-blocks.

8. The method according to any one of claims 1-7, characterized in that, The process of fusing multiple windowed block images based on a preset fusion strategy to obtain a fused image includes: Each windowed sub-block image is expanded outward to obtain an expanded sub-block image; wherein, there are overlapping regions between the expanded sub-block images; For every two adjacent expanded sub-block images, determine two target window functions corresponding to the two expanded sub-block images; Based on the preset fusion strategy and the function types corresponding to the two target window functions, the target fusion method of the two adjacent expanded sub-block images is determined. Based on the fusion method corresponding to each adjacent expanded sub-block image, the overlapping regions are fused to obtain the fused image.

9. A near-field MIMO radar sidelobe suppression device based on phase gradient self-focusing and spatial adaptive time-frequency window function, characterized in that, include: The acquisition module is used to acquire MIMO radar echo signals of the target scene and generate a wavenumber domain image based on the MIMO radar echo signals. The first processing module is used to perform phase gradient autofocusing processing on the wavenumber domain image to obtain a phase-corrected image; the phase gradient autofocusing is used to eliminate phase deviation in the wavenumber domain image and improve the clarity of the image boundary. The second processing module is used to perform spatial adaptive partitioning based on the phase-corrected image to obtain multiple sub-block images; The third processing module is used to perform window function matching for each sub-block image, determine the target window function corresponding to each sub-block image, and window the sub-block image based on the target window function to obtain a windowed sub-block image; The window function is used to perform weighted calculations on the sub-block image to improve the clarity of the sub-block image boundaries; The fourth processing module is used to fuse multiple windowed block images based on a preset fusion strategy to obtain a fused image; The fifth processing module is used to perform secondary phase gradient autofocus processing on the fused image to obtain the target imaging result.

10. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-8.

12. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-8.

Images