Radar high-resolution imaging method and system based on dynamic windowing processing

Abstract

The application belongs to the technical field of radar high-resolution imaging, and particularly discloses a radar high-resolution imaging method and system based on dynamic windowing processing, wherein the method comprises the following steps: acquiring original complex echo data, generating a pre-detection complex range image according to the original complex echo data, determining a noise region, a strong scattering region, a weak scattering region and a protection region according to the pre-detection complex range image, and performing distance high-resolution branch processing and sidelobe suppression branch processing on the original complex echo data; and the technical problem solved by the application is how to control the distance main lobe width and the sidelobe amplitude, screen the pulses that can be used for coherent accumulation, respectively complete the distance migration correction and focusing of the stationary scattering unit and the moving scattering unit, and reduce the repeated calculation caused by the processing parameter adjustment under the condition that the strong scattering unit and the weak scattering unit coexist, the effective pulse phase communication relationship is unstable, and the stationary scattering unit and the moving scattering unit have different distance trajectories and phase histories.

Description

Technical Field

[0001] This invention relates to the field of high-resolution radar imaging technology, and more specifically to a high-resolution radar imaging method and system based on dynamic windowing processing. Background Technology

[0002] Radar echoes exhibit uneven scattering intensity distribution along the range direction, with strong scattering elements' sidelobes obscuring adjacent weak scattering elements, and noise amplitude varying with range position. When processing all range sampling points using the same range window, it's difficult to simultaneously control the range main lobe width and range sidelobe amplitude. The complex results from two branches obtained using different range windows may also have differences in the position and complex phase of the main peak. Direct fusion without correction can lead to main peak misalignment, complex cancellation, or discontinuities at region boundaries. Radar platform motion errors, target motion, and local signal-to-noise ratio variations can cause some pulses to lack stable phase relationships. If all continuously acquired pulses are directly used for coherent accumulation, low-correlation pulses will affect azimuth focusing; simply truncating the aperture by a fixed number may reduce the usable aperture time span. Stationary and moving scattering elements have different range trajectories and phase histories. Using the same range migration correction and focusing parameters may result in residual range migration, main lobe broadening, or positional deviation. Different processing parameters apply to local distance processing, phase connectivity aperture construction, distance migration correction, and branch focusing, respectively. If the entire imaging process is re-executed every time the parameters are adjusted, unaffected processing steps will be recalculated. Therefore, it is necessary to clearly define the input, output, and dependent parameters of each processing result to establish an executable data processing relationship between local distance processing, phase connectivity pulse selection, stationary or moving focusing, image quality verification, and regional output. Summary of the Invention

[0003] This invention provides a radar high-resolution imaging method and system based on dynamic windowing processing. The technical problem solved by this invention is how to control the range main lobe width and side lobe amplitude under the conditions of coexistence of strong and weak scattering units, unstable effective pulse phase connectivity, and different range trajectories and phase histories of stationary and moving scattering units, to select pulses that can be used for coherent accumulation, to complete range migration correction and focusing of stationary and moving scattering units respectively, and to reduce redundant calculations caused by processing parameter adjustments.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A radar high-resolution imaging method based on dynamic windowing processing includes: acquiring raw complex echo data; generating a pre-detection complex range image based on the raw complex echo data; determining noise regions, strong scattering regions, weak scattering regions, and protection regions based on the pre-detection complex range image; performing range high-resolution branching and sidelobe suppression branching processing on the raw complex echo data; and obtaining locally range-processed complex echoes through group delay correction, phase correction, and dynamic fusion; determining target connectivity regions and phase continuity segments based on the locally range-processed complex echoes; establishing a phase connectivity aperture map; and extracting target-level phase connectivity pulses corresponding to each target connectivity region from the phase connectivity aperture map. The system first obtains a set of target-level phase-connected pulse indexes; then, based on this set, performs range migration correction and stationary or moving branch focusing on the local range-processed complex echoes to generate a region parameter version complex image set, and establishes a parameter dependency graph between each processing parameter and the region parameter version complex image set; it then verifies the region parameter version complex image set according to the imaging quality qualification conditions, maps the quality items that fail the verification to the corresponding processing parameter set, performs incremental recalculation on the affected processing nodes according to the parameter dependency graph, determines the region output results from the verified parameter version complex images, and generates the final high-resolution complex image based on each region output result.

[0005] In one aspect of the invention, determining the noise region, the strong scattering region, the weak scattering region, and the protection region includes: A fixed reference distance window is used to perform pre-detection range compression on the original complex echo data to obtain the pre-detection complex range image, and the pre-detection complex range image is divided into overlapping local range processing output intervals along the range direction; The noise region determination threshold is determined based on the amplitude statistics of the targetless distance interval, and the strong scattering region determination threshold above the noise region determination threshold is determined based on the distance compression of the main peak amplitude of the calibrated strong scattering unit. Distance sampling points with amplitudes not lower than the strong scattering region determination threshold are defined as the strong scattering region. Distance sampling points located within the range of the distance sidelobes on both sides of the strong scattering region but not belonging to the strong scattering region are defined as the protection region. Among the remaining distance sampling points, distance sampling points with amplitudes not higher than the noise region determination threshold are defined as the noise region. The remaining distance sampling points are defined as the weak scattering region.

[0006] In one aspect of the invention, obtaining the local distance processing complex echo includes: For each of the local distance processing output intervals, an extended input interval covering the effective response length of the distance matched filter is extracted from the original complex echo data. Two distance matched filters are constructed using a first distance window and a second distance window with a higher degree of apodization than the first distance window. High-resolution distance branch processing and sidelobe suppression branch processing are performed on the extended input intervals. Output sampling points affected by convolution boundaries are deleted, and the complex processing results of the two branches corresponding to the local distance processing output intervals are retained. Based on the distance to the main peak position difference and the complex phase difference of the main peak after the calibration scattering unit is processed by two branches, group delay correction and phase correction are performed on the complex processing results of the two branches. Dynamic fusion weights are determined based on the regions to which each distance sampling point belongs. The complex processing results of the two branches after correction are fused point by point, and the overlapping parts of adjacent local distance processing output intervals are spliced ​​in the complex domain.

[0007] In one aspect of the invention, determining the target connected region and the phase continuity segment includes: The effective detection threshold is determined based on the local noise amplitude of the local distance processed echo, and the distance sampling points whose amplitude reaches the effective detection threshold and are continuous in the distance direction and slow time direction are merged into the target connected region. In each target connected region, the effective distance sampling points in adjacent pulses are aligned with the distance trajectory, the degree of complex correlation between adjacent pulses is calculated, and the predicted phase difference corresponding to the platform motion geometry and the actual distance trajectory is subtracted from the observed phase difference to obtain the absolute value of the remaining phase difference between adjacent pulses. Continuous pulses whose complex correlation reaches the phase continuity determination threshold and whose absolute value of the remaining phase difference between adjacent pulses does not exceed the phase change determination threshold are classified into the same phase continuity segment.

[0008] In one aspect of the invention, establishing the phase connectivity aperture map includes: Each of the phase continuity segments is set as a node in the phase connectivity aperture diagram; When the pulse ranges of two phase continuous segments belonging to the same target connected region are adjacent, or the number of interrupted pulses between the two pulse ranges does not exceed the allowed number of interruptions, the comparison pulses at the boundary of the two phase continuous segments are extracted respectively, the effective distance sampling points in the comparison pulses are aligned with the distance trajectory, and the degree of complex correlation between segments and the absolute value of the remaining phase difference between segments are determined based on the aligned complex echo. When the complex correlation between the segments reaches the segment connectivity determination threshold and the absolute value of the remaining phase difference between the segments does not exceed the segment phase error threshold, a connection edge is established between the corresponding nodes.

[0009] In one aspect of the invention, extracting the target-level phase-connected pulse index set includes: In the phase connectivity aperture diagram corresponding to each target connectivity region, find the set of connected nodes and delete the set of connected nodes whose effective pulse count does not reach the minimum coherent accumulation pulse count or whose aperture time span does not reach the minimum aperture time span. The target set of connected nodes is determined according to the aperture time span, number of effective pulses, average complex correlation degree, absolute value of the remaining phase difference between the largest adjacent pulses, and number of pulse interruptions of the other connected node sets, in a preset comparison order. The pulse numbers covered by the target connected node set are arranged according to the acquisition time, and a validity weight is recorded for each pulse number to obtain the target-level phase connected pulse index set.

[0010] In one aspect of the invention, generating the complex image group of the region parameter versions and establishing the parameter dependency graph includes: Based on the maximum distance deviation between the actual distance trajectory and the stationary distance trajectory of the target connected region and the remaining phase error after deducting the stationary predicted phase, the target connected region is divided into stationary scattering units or moving scattering units, and distance migration correction is performed according to the corresponding target-level phase connected pulse index set. Static branch focusing is performed on the static scattering unit, and motion branch focusing is performed on the moving scattering unit according to a finite set of candidate motion parameters to obtain multiple parameter versions of complex images of the same target connected region; Record the local distance processing parameters, target-level phase connectivity pulse index set, distance migration correction parameters, and branch focusing parameters that each parameter version complex image depends on, and establish the parameter dependency graph according to the input-output relationship of the processing results.

[0011] In one aspect of the invention, performing incremental recalculation and generating the final high-resolution complex image includes: The imaging quality within a region is qualified by using the distance main lobe width, azimuth main lobe width, peak sidelobe ratio, integral sidelobe ratio, and remaining range migration. The quality of stitching between regions is qualified by using the degree of complex correlation in the overlapping region and the absolute value of the remaining phase error in the overlapping region. Map the range main lobe width, peak sidelobe ratio, or integral sidelobe ratio that failed the verification to the range window and dynamic fusion weight processing parameter group; map the phase consistency quality items that failed the verification to the phase consistency processing parameter group; map the remaining range migration that failed the verification to the range migration correction processing parameter group; and map the azimuth main lobe width that failed the verification to the branch focusing processing parameter group. The downstream processing node pointed to by the corresponding processing parameter group is determined according to the parameter dependency graph. Only the downstream processing node is recalculated, and the output result of the region that passes the imaging quality qualification condition and the inter-region stitching quality qualification condition is written into the unified imaging coordinate system.

[0012] In another aspect, the present invention also relates to a high-resolution radar imaging system based on dynamic windowing processing, comprising: The echo local range processing module is used to acquire raw complex echo data, generate pre-detection complex range images, identify noise regions, strong scattering regions, weak scattering regions and protection regions, perform high-resolution range branch processing and sidelobe suppression branch processing, and obtain locally range processed complex echoes through group delay correction, phase correction and dynamic fusion. The phase connectivity aperture determination module is used to determine the target connectivity region and phase continuity segment based on the local distance processed complex echo, establish a phase connectivity aperture map, and extract the target-level phase connectivity pulse index set; The branch focusing and version generation module is used to perform distance migration correction and static or moving branch focusing based on the target-level phase connectivity pulse index set, generate a complex image group of region parameter versions and establish a parameter dependency graph. The quality verification and image output module is used to verify the complex image group of the region parameter version according to the imaging quality qualification conditions, map the quality items that fail the verification to the corresponding processing parameter group, perform incremental recalculation according to the parameter dependency graph, and generate the final high-resolution complex image based on the output results of the regions that pass the verification.

[0013] Compared with the prior art, the present invention has the following beneficial effects: This invention first generates a pre-detection complex range image using a fixed reference range window. Then, it determines noise regions, strong scattering regions, weak scattering regions, and protection regions based on the amplitude of the pre-detection complex range image and the sidelobe coverage of strong scattering units, making region classification independent of the yet-to-be-generated dynamic fusion results. The high-resolution range branch and the sidelobe suppression branch use extended input data supported by complete matched filtering. After group delay correction and phase correction are completed for the complex results of the two branches, complex fusion is performed according to the region type, which can reduce the misalignment of the main peaks and complex cancellation between the two branches. Phase continuity segments and phase connectivity aperture maps are used to record the complex correlation relationships and residual phase errors between different pulses in the target connected region. The target-level phase connectivity pulse index set simultaneously records the pulse number and validity weight, allowing the pulses used for coherent accumulation and their aperture time positions to be determined. The stationary branch and the moving branch perform range migration correction, propagation phase compensation, and complex accumulation based on the stationary range trajectory and the candidate moving range trajectory, respectively, so that scattering units in different motion states use corresponding processing parameters. The parameter dependency graph uses specific processing objects and parameter versions as node attributes. When a quality item fails to meet the requirements, only the downstream processing results of the corresponding processing parameter node are recalculated, while unaffected intermediate results are preserved. The final image also needs to pass regional imaging quality compliance conditions and inter-region stitching quality compliance conditions to ensure that the regional output results have corresponding processing parameters and quality verification records. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the radar high-resolution imaging method based on dynamic windowing processing according to the present invention.

[0016] Figure 2 This is a flowchart of a sub-step in step 1 of the radar high-resolution imaging method based on dynamic windowing processing of the present invention.

[0017] Figure 3 This is a flowchart of a sub-step in step 2 of the radar high-resolution imaging method based on dynamic windowing processing of the present invention.

[0018] Figure 4 This is a flowchart of the sub-step of step 3 in the radar high-resolution imaging method based on dynamic windowing processing of the present invention.

[0019] Figure 5This is a flowchart of the sub-step 4 in the radar high-resolution imaging method based on dynamic windowing processing of the present invention.

[0020] Figure 6 This is a framework diagram of the radar high-resolution imaging system based on dynamic windowing processing according to the present invention.

[0021] Figure 7 This is a simplified structural diagram of the radar high-resolution imaging system based on dynamic windowing processing according to the present invention.

[0022] Figure 8 This is a timing diagram of the radar high-resolution imaging system based on dynamic windowing processing according to the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to embodiments, which are used to illustrate the processing procedure of the present invention. Those skilled in the art can calibrate the corresponding parameters according to the radar operating mode while maintaining the input-output relationships of each processing stage.

[0024] like Figure 1 As shown, this embodiment discloses a radar high-resolution imaging method and system based on dynamic windowing processing, wherein the method specifically includes: S1. Obtain the original complex echo data, generate a pre-detection complex range image using a fixed reference range window, determine the noise region, strong scattering region, weak scattering region and guard region based on the pre-detection complex range image, perform range high-resolution branch processing and sidelobe suppression branch processing on the original complex echo data, and obtain the local range processed complex echo after group delay correction, phase correction and dynamic fusion. S2. Based on the local distance processing of the complex echo, determine the target connected region and phase continuous segment, establish a phase connectivity aperture map, and extract the target-level phase connectivity pulse index set corresponding to each target connected region from the phase connectivity aperture map; S3. Based on the target-level phase-connected pulse index set, perform range migration correction and stationary or moving branch focusing on the local range-processed complex echoes to generate a region parameter version complex image group, and establish a parameter dependency graph between each processing parameter and the region parameter version complex image group. S4. Based on the imaging quality qualification conditions, verify the region parameter version complex image group, map the quality items that fail verification to the corresponding processing parameter group, perform incremental recalculation on the affected processing nodes according to the parameter dependency graph, determine the region output result from the parameter version complex image that passes verification, and generate the final high-resolution complex image.

[0025] like Figure 2 As shown, in this embodiment, S1 includes: S11. Perform integrity verification on the original complex echo data, generate a pre-detection complex range image using a fixed reference distance window, divide the pre-detection complex range image into overlapping local range processing output intervals, and determine the noise region, strong scattering region, weak scattering region, and protection region. S12. Extract extended input data with complete range matched filtering support for each local range processing output interval, perform range high-resolution branch processing and sidelobe suppression branch processing respectively, and perform group delay correction and phase correction on the complex processing results of the two branches. S13. Determine the dynamic fusion weights based on the regions to which each distance sampling point belongs, fuse the two branches of complex processing results point by point, and perform complex domain splicing on the overlapping parts of adjacent local distance processing output intervals to obtain local distance processing complex echoes.

[0026] like Figure 3 As shown, in this embodiment, S2 includes: S21. Determine the target connected region based on the local distance processing complex echo, perform distance trajectory alignment on adjacent pulses within each target connected region, calculate the complex correlation degree between adjacent pulses and the absolute value of the remaining phase difference between adjacent pulses, and obtain a phase continuous segment. S22. Using phase continuous segments as nodes, establish connection edges based on the degree of complex correlation between segments and the absolute value of the remaining phase difference between segments to obtain the phase connectivity aperture map corresponding to each target connected region. S23. Select the set of target connected nodes that meet the requirements of effective pulse quantity and aperture time span from the phase connectivity aperture diagram. Arrange the pulse numbers covered by the target connected node set according to the acquisition time and record the effectiveness weight to obtain the target-level phase connectivity pulse index set.

[0027] like Figure 4 As shown, in this embodiment, S3 includes: S31. Based on the actual distance trajectory, stationary distance trajectory and remaining phase error of the target connected region, divide the target connected region into stationary scattering units or moving scattering units, and perform range migration correction based on the target-level phase connected pulse index set. S32. Perform static branch focusing on the stationary scattering unit and perform motion branch focusing on the moving scattering unit according to a finite set of candidate motion parameters to generate multiple parameter version complex images corresponding to the connected region of the same target, and obtain a group of region parameter version complex images. S33. Record the local distance processing parameters, target-level phase connectivity pulse index set, distance migration correction parameters, and branch focusing parameters used in the complex image of each parameter version, and establish a parameter dependency graph according to the input-output relationship of the intermediate processing results.

[0028] like Figure 5 As shown, in this embodiment, S4 includes: S41. Using the range main lobe width, azimuth main lobe width, peak side lobe ratio, integral side lobe ratio and remaining range migration to form the imaging quality qualification conditions within the region, the region parameter version complex image group is checked within the region. S42. Map the quality items that failed the verification to the corresponding processing parameter group, determine the downstream processing node of the corresponding processing parameter node according to the parameter dependency graph, and perform incremental recalculation only on the downstream processing node. S43. Determine the candidate region output results of each target connected region from the parameter version complex image that has passed the imaging quality qualification condition within the region, perform phase alignment and complex domain stitching on the output results of adjacent candidate regions, and generate the final high-resolution complex image after the inter-region stitching quality qualification condition is checked.

[0029] In this embodiment, the original complex echo data is two-dimensional complex sampling data formed by the radar receiver after orthogonal demodulation and analog-to-digital conversion. One dimension corresponds to the range sampling order, and the other dimension corresponds to the pulse acquisition order. Radar control parameters include carrier frequency, signal bandwidth, pulse width, range sampling frequency, pulse repetition frequency, and number of pulse acquisitions. Platform navigation parameters include the platform position, platform velocity, and platform attitude quaternions at each pulse acquisition time. The echo local range processing module checks the number of pulses, the number of range sampling points per pulse, the echo timestamp, and the navigation timestamp in the original complex echo data. For a missing single range sampling point, linear interpolation is performed on the real and imaginary parts of adjacent valid range sampling points. For intervals with consecutive missing values ​​exceeding the number of sampling points occupied by the range compression main lobe width, these intervals are marked as invalid range intervals, and the interpolation results are not used in subsequent processing.

[0030] In some embodiments, for pulses with missing navigation parameters, position and velocity are linearly interpolated according to the time ratio between the missing pulse timestamp and two consecutive valid navigation moments. Attitude quaternions are interpolated using spherical interpolation, and the interpolation results are normalized. The allowable number of navigation calibrations is determined based on the correspondence between the number of consecutively missing navigation parameters and the interpolation error. During calibration, navigation parameters for 1 to 5 consecutive pulses are deleted from the flight record with complete navigation data, and the reconstructed result is compared with the original navigation result. In this embodiment, the maximum position errors for 1 to 5 consecutively missing pulses are 0.012m, 0.026m, 0.047m, 0.071m, and 0.104m, respectively. 10% of the theoretical resolution in the range direction is 0.05m; therefore, the allowable number of navigation calibrations is determined to be 3 consecutive pulses. When more than 3 consecutive navigation parameters are missing, the corresponding pulses are marked as invalid pulses and are not included in coherent accumulation as valid pulses.

[0031] In this embodiment, the echo local range processing module uses a rectangular reference range window to construct a fixed reference range matched filter. It performs pre-detection range compression on the complete range of each pulse against the original complex echo, obtaining a pre-detection complex range image. This pre-detection complex range image is only used to determine the region type and is not used as the final processing result of the range high-resolution branch or sidelobe suppression branch. This allows the region type to be determined before the two dynamic windowing branch processing steps.

[0032] In this embodiment, the local distance processing output range represents the range of distance sampling points ultimately retained after the two branches complete distance compression. When performing two-branch processing, the echo local distance processing module extracts extended input data from the original complex echo that covers the complete and effective response of the range-matched filter. It performs linear convolution using an overlap-preservation method, deletes sampling points affected by the convolution boundary, and retains only the complex processing results corresponding to the local distance processing output range. Zero-padding is applied to portions located at the edge of the complete distance record where complete extended input data cannot be obtained. The output sampling points corresponding to the zero-padding are not used for noise statistics, target detection, or image quality measurement.

[0033] In this embodiment, the noise region determination threshold is obtained based on the pre-detected complex range image without obvious targets. During the calibration phase, a distance interval excluding echoes from corner reflectors, buildings, and vehicles is selected. The normalized amplitude of each distance sampling point is statistically analyzed, and the 95th percentile of all noise samples is taken as the noise region determination threshold. The strong scattering region determination threshold is obtained based on the pre-detected distance compression of the main peak amplitude of a stationary corner reflector. The processor normalizes the distance propagation loss of the main peak amplitude of the corner reflector under different observation distances and azimuth angles, and takes the lower limit of the normalized main peak amplitude as the strong scattering region determination threshold, ensuring that the strong scattering region determination threshold is higher than the noise region determination threshold.

[0034] In one example, 30 groups of echoes without obvious targets provided 192,000 noise amplitude samples. The 95th percentile of the normalized amplitude was 0.018, therefore the noise region determination threshold was set to 0.018. The lower limit of the main peak amplitude of the 30 groups of stationary corner reflector echoes after distance propagation loss normalization was 0.42, therefore the strong scattering region determination threshold was set to 0.42. The echo local distance processing module first marked sampling points with amplitudes not lower than 0.42 as strong scattering regions, then determined the protection zones on both sides of the strong scattering regions. Subsequently, it marked unmarked sampling points with amplitudes not higher than 0.018 as noise regions, and marked the remaining sampling points as weak scattering regions. Each sampling position determined by the distance sampling point number and pulse number has only one main region type.

[0035] In this embodiment, the protected region represents the range of adjacent range sampling points that may be affected by the range sidelobes of the strong scattering region. The processor obtains the range compression response after high-resolution range branching from the stationary corner reflector calibration data. The position where the amplitude of the sidelobes on both sides of the corner reflector's main peak decreases to the noise region determination threshold is taken as the sidelobe coverage boundary. The number of range sampling points between this boundary and the boundary of the strong scattering region is counted, and the maximum statistical value under different calibration distances and different signal-to-noise ratios is taken as the width of the protected region. The protected region does not include the strong scattering region itself. The protected region located at the edge of the local range processing output interval extends to the adjacent local range processing output interval according to the global range sampling point number.

[0036] In some embodiments, when multiple strong scattering regions exist within the same local distance processing output interval, overlapping guard regions are merged into a continuous guard region. Sampling points within a guard region are no longer simultaneously labeled as weak scattering regions or noise regions, but one or more adjacent strong scattering region identifiers can be recorded. These adjacent strong scattering region identifiers are used to determine the dynamic fusion weight of the sampling point without changing the main region type of that sampling point.

[0037] In this embodiment, the high-resolution range branch and the sidelobe suppression branch receive the same extended input data, the same transmitted signal reference data, and the same output sampling point range. Both branches first perform a range-to-frequency domain transform on the extended input data, then multiply it by the complex conjugate of the transmitted signal reference spectrum with the first and second range windows applied, respectively. Subsequently, an inverse range-to-frequency transform is performed to remove sampling points affected by linear convolution boundaries, retaining the complex range compression result corresponding to the local range processing output interval. The first and second range windows have the same length, the same center of symmetry, and the same normalization method, with the second range window exhibiting a higher degree of apodization than the first range window.

[0038] In this embodiment, the first and second distance windows adopt standard Kaiser windows, and the shape parameters of a finite number of Kaiser windows are verified one by one during the calibration phase. The first distance window selects the candidate parameter with the smallest distance from the main lobe width from the candidate parameters whose distance from the main lobe width does not exceed 1.2 times the theoretical main lobe width; the second distance window selects the parameter with the smallest distance from the main lobe width from the candidate parameters whose peak sidelobe ratio does not exceed 22dB and integral sidelobe ratio not higher than The candidate parameters for 12dB are selected based on the minimum distance from the main lobe width. In a calibration example, when the Kaiser window shape parameters are 0, 2, 3, and 4, the distances from the main lobe width are 0.50m, 0.55m, 0.59m, and 0.65m, respectively, and the peak sidelobe ratios are respectively... 13.2dB 18.8dB 24.1dB and 30.2dB, integral sidelobe ratios are respectively 9.8dB 11.1dB 13.0dB and 15.5dB, therefore the Kaiser window shape parameter of the first distance window is set to 0, and the Kaiser window shape parameter of the second distance window is set to 3.

[0039] In this embodiment, using different distance windows for the two branches may result in a difference in the position of the main peak and a difference in the complex phase of the main peak. The echo local distance processing module uses the high-resolution distance branch as the correction reference, inputting the original complex echoes from the same calibrated corner reflector into the two branches respectively, and recording the difference in the position of the main peak of the sidelobe suppression branch relative to the high-resolution distance branch. The integer part of the difference in the position of the main peak of the distance is corrected by moving the distance sampling sequence of the sidelobe suppression branch, and the fractional part is corrected by applying a linear phase change corresponding to the fractional distance delay in the distance frequency domain.

[0040] In this embodiment, after group delay correction is completed, the processor calculates the sum of the products of the two branch complex echoes at the main peak of the corner reflector and its adjacent effective sampling points. The phase of the sum is taken as the complex phase difference of the main peak, and a complex phase rotation in opposite directions is applied to all complex processing results of the sidelobe suppression branch. The correction parameters are stored separately according to the combination of carrier frequency, signal bandwidth, range sampling frequency, first range window parameters, and second range window parameters, and are not directly reused between different radar operating modes.

[0041] In one calibration example, the median distance difference between the two branches corresponding to the main peak positions of 30 sets of corner reflector echoes is 0.37 distance sampling points, and the complex phase difference of the main peak is 12.4°. After group delay correction and phase correction, the remaining distance difference between the main peak positions is 0.04 distance sampling points, and the remaining complex phase difference of the main peak is 1.6°. The allowable error for group delay correction is 0.1 distance sampling points, and the allowable error for phase correction is 5°. The condition for determining these errors is that the decrease in the amplitude of the main peak of the corner reflector after complex fusion of the two branches relative to the amplitude of the larger main peak of the two branches does not exceed 0.2 dB. If the remaining error exceeds either allowable error, dynamic fusion is not performed, and the corresponding range window parameter combination is marked as failing correction.

[0042] In this embodiment, the dynamic fusion weights are determined from a finite candidate set of 0, 0.25, 0.5, 0.75, and 1 during the calibration phase. The processor uses each candidate weight to perform complex fusion on the joint echoes from the strong and weak angle reflectors, measuring the distance main lobe width, peak sidelobe ratio, and integral sidelobe ratio. For the strong scattering region and the guard region, the dynamic fusion weight of the smallest sidelobe suppression branch that simultaneously satisfies the three constraints is selected; for the noisy region and the weak scattering region unaffected by strong scattering sidelobes, the dynamic fusion weight of the smallest distance main lobe width without producing additional amplitude abrupt changes at the boundary is selected.

[0043] In one calibration example, when the dynamic fusion weight of the sidelobe suppression branch is set to 0.5, the peak sidelobe ratio is: 20.8dB, not reached The limit value is 22dB; when the sidelobe suppression branch dynamic fusion weight is 0.75, the distance from the main lobe width is 0.58m, and the peak sidelobe ratio is... 22.6dB, integral sidelobe ratio is The sidelobe suppression branch dynamic fusion weight is set to 0.75 for both the strong scattering region and the protected region, reaching 12.4dB, which meets the corresponding limit. For the noisy region and the weak scattering region far from the strong scattering region, the high-resolution range branch processing result is used, and the sidelobe suppression branch dynamic fusion weight is set to 0.

[0044] The results of the complex number processing of the two branches are merged point by point according to the following formula:

[0045] in, Indicates the distance from the sampling point In slow time Local distance processing of the echo at the location; Indicates the distance from the sampling point In slow time The complex number processing result after sidelobe suppression branching, group delay correction, and phase correction; Indicates the distance from the sampling point In slow time The complex processing result after high-resolution branching of the distance; Indicates the distance from the sampling point In slow time The sidelobe suppression branch dynamically fuses weights at the specified location, with values ​​ranging from the closed interval. ; Indicates the global distance sampling point number; This indicates the slow time corresponding to the pulse acquisition moment.

[0046] In this embodiment, eight weight transition zones are set on both sides of the boundary of each region, corresponding to the distance from the sampling points. The dynamic fusion weight of the sidelobe suppression branch in the transition zone varies equally from the determined value of one side region to the determined value of the other side region according to the number of sampling points. For a protected region that is simultaneously affected by the sidelobes of multiple strong scattering regions, candidate weights are determined according to each adjacent strong scattering region, and the maximum value among them is used as the dynamic fusion weight of the sidelobe suppression branch for that sampling point.

[0047] In this embodiment, the overlapping portions of adjacent local distance processing output intervals are stitched together in the complex domain using linear stitching weights. The stitching weight of the preceding local distance processing output interval decreases from 1 to 0 point by point, while the stitching weight of the following local distance processing output interval increases from 0 to 1 point by point. The sum of the two stitching weights at the same global distance sampling point is 1. Before stitching, the processor uses sampling points in the overlapping portion whose amplitude reaches three times the local noise amplitude to calculate the average complex phase difference between the two intervals and performs complex phase alignment on the following local distance processing output interval. When the overlapping portion has fewer than 16 effective sampling points, the overlapping portion is not used to estimate the phase difference, and the two intervals use the same calibration phase correction parameters. After stitching is completed, the output is a local distance processing complex echo that maintains the global distance sampling point order and pulse acquisition order, and the region type, dynamic fusion weight, local distance processing output interval identifier, and two branch processing parameters are associated and saved.

[0048] In this embodiment, the phase connectivity aperture determination module selects a noise region in each local distance processing output interval, determines the 95th percentile of the local distance processing complex echo amplitude within the noise region as the local noise amplitude of the current interval, and sets three times the local noise amplitude as the effective detection threshold. When there are fewer than 64 effective sampling points in the noise region within the current interval, the module extracts the local noise amplitudes corresponding to temporally adjacent effective pulses and geographically adjacent local distance processing output intervals, and takes the median value after sorting as the local noise amplitude of the current interval.

[0049] In one detection example, the 95th percentile of the noise region amplitude within the current local distance processing output interval is 0.017, therefore the effective detection threshold is set to 0.051. The 3-fold relationship was verified using 30 sets of echoes from targets without obvious targets and 30 sets of echoes from weakly scattering targets. When using 3 times the local noise amplitude, the proportion of noise sampling points misclassified as effective distance sampling points was 0.6%, while the retention rate of effective sampling points for weakly scattering targets was 94%. Therefore, 3 was determined as the effective detection threshold multiple.

[0050] In this embodiment, the phase connectivity aperture determination module marks sampling points whose amplitude reaches the effective detection threshold as effective distance sampling points, and performs connectivity search according to distance-oriented adjacency and slow-time adjacency. If the distance sampling points in the same pulse are adjacent in number, or if the distance position change in adjacent pulses does not exceed the allowable distance change, the corresponding effective distance sampling points are grouped into the same candidate connected region. The allowable distance change is determined based on the maximum platform displacement between adjacent pulses, the target's preset maximum radial displacement, and the upper limit of navigation position error.

[0051] In one example, the platform speed is 120m / s, the pulse repetition frequency is 1200Hz, and the platform displacement between adjacent pulses is 0.1m; the target's preset maximum radial velocity is 15m / s, and the target's radial displacement between adjacent pulses is 0.0125m; after adding a navigation position error upper limit of 0.05m, the maximum possible distance change is 0.1625m, which is less than the distance sampling interval of 0.4167m. Therefore, the distance change is allowed to take 1 distance sampling point.

[0052] In this embodiment, candidate connected regions with fewer than 16 continuous pulses and fewer than 64 effective range sampling points are deleted as isolated noise regions. During calibration, combined detection is performed using 8, 12, 16, and 20 continuous pulses and 32, 48, 64, and 80 effective range sampling points, respectively. When using 16 continuous pulses and 64 effective range sampling points, all isolated noise regions in the 30 groups without obvious target echoes are deleted, and 28 target regions in the 30 groups of weakly scattered target echoes are retained; therefore, the aforementioned values ​​are adopted.

[0053] In this embodiment, for the retained candidate connected regions, the processor multiplies the distance position of each effective distance sampling point by the corresponding echo power in each pulse, sums the products, and divides by the sum of the echo powers to obtain the amplitude centroid distance position of the pulse. The processor then forms the actual distance trajectory according to the pulse acquisition sequence. When the actual distance trajectory change of adjacent pulses exceeds the allowable distance change, the candidate connected region is split at the corresponding pulse boundary. When two candidate connected regions are adjacent in the distance direction, have the same direction of actual distance trajectory change, and the number of interrupt pulses does not exceed the allowable number of interrupts, the two candidate connected regions are merged.

[0054] In this embodiment, the allowed number of interruptions is determined by verifying each of the candidate values ​​from 0 to 8. The processor performs candidate region merging on 20 sets of low signal-to-noise ratio continuous moving target echoes and 20 sets of nearby dual-target echoes. When the allowed number of interruptions is 4, the continuous retention rate of the same target is 96%, and no incorrect merging of different targets occurs; when the allowed number of interruptions is 5, two cross-target mergings occur, therefore the allowed number of interruptions is determined to be 4 pulses. After splitting and merging, each target connected region records the target connected region identifier, global distance sampling point range, pulse number range, actual distance trajectory, and effective distance sampling point set.

[0055] In this embodiment, the phase connectivity aperture determination module calculates the complex correlation between adjacent pulses within each target connectivity region. Before calculation, the effective distance sampling points in the subsequent pulse are aligned to the relative distance positions of the preceding pulse according to the actual distance trajectory. Non-integer distance offsets are achieved using 8-point windowed sine interpolation. The processor multiplies the complex echo of the preceding pulse with the complex conjugate of the complex echo of the subsequent pulse point by point, accumulates all products, takes the modulus, and then divides by the non-negative square root of the product of the energy of the complex echoes of the two pulses to obtain the complex correlation between adjacent pulses.

[0056] In this embodiment, the processor obtains the phase of the aforementioned complex echo multiplication and accumulation result, gets the observed phase difference between adjacent pulses, then determines the predicted phase difference between adjacent pulses based on the platform antenna phase center position, carrier frequency, and the actual distance trajectory of the target, subtracts the predicted phase difference from the observed phase difference, and adjusts the resulting phase to... Within the range of 180° to 180°, the absolute value is taken as the absolute value of the remaining phase difference between adjacent pulses. When the number of common effective distance sampling points involved in the calculation is less than the number of minimum sampling points for phase detection, the corresponding pulse boundary is marked as an undecidable boundary, and a new phase continuity segment is divided at this boundary.

[0057] In this embodiment, the minimum number of sampling points for phase detection is 8, 12, 16, and 24 for repeated measurements, corresponding to maximum repeated measurement fluctuations of 0.091, 0.066, 0.043, and 0.029 for the complex correlation degree, respectively. Sixteen effective distance sampling points ensure that the repeated measurement fluctuation of the complex correlation degree does not exceed 0.05; therefore, the minimum number of sampling points for phase detection is 16. The phase continuity determination threshold is 0.85, and the phase change determination threshold is 45°. These thresholds are determined by the condition that after coherent accumulation using pulses that meet both thresholds, the amplitude loss of the main peak of the stationary corner reflector does not exceed 0.5 dB. When the complex correlation degree reaches 0.85 and the absolute value of the remaining phase difference between adjacent pulses does not exceed 45°, adjacent pulses are classified into the same phase continuity segment.

[0058] In this embodiment, each phase continuous segment records the target connected region identifier, start pulse number, end pulse number, effective distance sampling point set, average complex correlation degree, and absolute value of the maximum adjacent pulse remaining phase difference. The phase connectivity aperture determination module sets each phase continuous segment as a node in the phase connectivity aperture graph. Phase continuous segments belonging to different target connected regions do not establish connection edges. Two phase continuous segments belonging to the same target connected region enter the segment connectivity determination stage if their pulse ranges are adjacent or the number of interrupted pulses does not exceed 4.

[0059] In this embodiment, when performing segment connectivity determination, 16 consecutive valid pulses are extracted from the end of the previous phase continuous segment and the beginning of the next phase continuous segment as boundary comparison pulses; if any segment has fewer than 16 valid pulses, all valid pulses of that segment are extracted, but not less than 8. The processor aligns the valid distance sampling points in the boundary comparison pulses to the same relative distance position according to the actual distance trajectories of the two segments, and then establishes comparison data pairs according to the time order of the pulses relative to the segment boundary.

[0060] In this embodiment, the degree of complex correlation between segments is determined using the same complex multiplication accumulation and energy normalization method as the degree of complex correlation between adjacent pulses. The processor obtains the phase of the complex echo multiplication accumulation result of the comparison data pair, subtracts the predicted phase changes corresponding to the platform motion geometry and the actual distance trajectory of the target, and adjusts the obtained phase to... The absolute value of the remaining phase difference between segments is obtained by taking the absolute value within the range of 180° to 180°. The segment connectivity threshold is set to 0.85, and the segment phase error threshold is set to 45°. A connection edge is established between two nodes when the complex correlation between segments reaches 0.85, the absolute value of the remaining phase difference between segments does not exceed 45°, and the number of interrupt pulses does not exceed 4. No connection edge is established when the number of valid comparison data pairs is less than 16.

[0061] In some embodiments, when multiple non-connected node sets are formed in the same target connected region, each node set and its pulse coverage range are stored separately. The phase connectivity aperture diagram is stored using an adjacency list. Each node records the phase continuity segment attributes, and each connecting edge records the degree of complex correlation between segments, the absolute value of the remaining phase difference between segments, the number of interrupted pulses, and the number of valid comparison data pairs participating in the determination.

[0062] In this embodiment, the phase connectivity aperture determination module performs a connectivity search on the phase connectivity aperture map corresponding to each target connected region, obtaining one or more sets of connected nodes. For each set of connected nodes, the number of different valid pulse numbers, the aperture time span between the first and last pulses, the average complex correlation degree, the absolute value of the remaining phase difference between the largest adjacent pulses, and the number of pulse interruptions are counted. If the number of valid pulses in a set of connected nodes does not reach the minimum number of coherent accumulated pulses, or the aperture time span does not reach the minimum aperture time span, the set of connected nodes is deleted.

[0063] In this embodiment, the minimum coherent accumulation pulse count and the minimum aperture time span are jointly determined using the azimuth resolution calibration data of the corner reflector. During calibration, the number of effective pulses and the time span between the first and last effective pulses are varied, and the amplitude of the corner reflector's main peak and the width of its azimuth main lobe are measured. In one calibration example, under the conditions of a 9.6 GHz carrier frequency, a center slant distance of 10,000 m, and a platform velocity of 120 m / s, when the number of effective pulses is not less than 480 and the aperture time span is not less than 1.67 s, the amplitude of the corner reflector's main peak reaches the specified value, and the width of its azimuth main lobe does not exceed 1.2 times the corresponding theoretical width of its azimuth main lobe. Therefore, the minimum coherent accumulation pulse count is taken as 480, and the minimum aperture time span is taken as 1.67 s.

[0064] In this embodiment, for the remaining connected node sets, the processor first selects the connected node set with the larger aperture time span; if the aperture time spans are the same, it selects the connected node set with the larger number of effective pulses; if the number of effective pulses is the same, it selects the connected node set with the higher average complex correlation; if the average complex correlation is the same, it selects the connected node set with the smaller absolute value of the remaining phase difference between the largest adjacent pulses; if the above items are still the same, it selects the connected node set with the fewer pulse interruptions. The processor determines the target connected node set according to this fixed comparison order.

[0065] In this embodiment, the target-level phase connectivity pulse index set is stored in a record table arranged by pulse acquisition time. Each record includes a pulse number and a validity weight. Pulses that pass phase connectivity determination and have valid distance sampling points are recorded with a weight of 1. Pulses located between two phase continuity segments with established connections but lacking valid distance sampling points due to their amplitude temporarily falling below the valid detection threshold retain their pulse numbers and are recorded with a weight of 0. Pulses marked as invalid due to continuous missing navigation parameters exceeding the allowed number for navigation calibration are not included in the record table. Distance migration correction retains all time positions in the record table, and complex coherent accumulation is performed only on pulses with a weight of 1 for stationary or moving branches.

[0066] In this embodiment, the branch focusing and version generation module extracts platform navigation parameters based on the pulse numbers in the target-level phase connectivity pulse index set, and converts the platform position into the radar antenna phase center position at each pulse acquisition time using radar antenna phase center calibration parameters. At the peak pulse time of the target connectivity region, the processor determines the initial spatial position of the target connectivity region based on the radar antenna phase center position, antenna beam pointing, slant range corresponding to the target amplitude centroid, and the reference elevation of the observation area. The reference elevation is provided by the imaging mission parameters; if no reference elevation is provided, the elevation of the center point of the observation area is used.

[0067] In this embodiment, a stationary range trajectory of the target in a stationary state is formed based on the distance between the radar antenna phase center position and the initial spatial position of the target's connected region at each pulse acquisition time. The actual range trajectory and the stationary range trajectory use the same pulse number, slant range starting point, and range sampling interval. The processor calculates the maximum range deviation between the two trajectories and subtracts the predicted phase change corresponding to the stationary range trajectory from the observed phase change of the target's connected region to obtain the absolute value of the remaining phase error in the stationary state.

[0068] In this embodiment, the static state determination threshold is determined through joint calibration of stationary corner reflectors and controlled moving vehicles. In one calibration example, the static state determination threshold is set to 0.5, 1, 1.5, and 2 distance sampling points, respectively. When using 1 distance sampling point, all 30 groups of stationary corner reflectors enter the stationary branch, and all 20 groups of controlled moving vehicles with radial displacement exceeding 1 distance sampling point enter the moving branch. Therefore, the static state determination threshold is set to 1 distance sampling point. When the maximum distance deviation does not exceed 1 distance sampling point and the absolute value of the remaining phase error in the static state does not exceed 45°, the target connected region is determined as a stationary scattering unit; if either condition is not met, the target connected region is determined as a moving scattering unit.

[0069] In this embodiment, for the stationary scattering unit, the processor determines the range position correction for each pulse based on the stationary range trajectory. For the moving scattering unit, the processor uses the center time of the target-level phase-connected pulse index set covering the aperture as the motion parameter reference time. The processor arranges the distance differences between the actual range trajectory and the stationary range trajectory at each effective pulse in chronological order, and obtains the initial radial velocity and initial radial acceleration through minimum residual fitting including first and second time terms. The radial velocity represents the first-order rate of change of the target slant range at the motion parameter reference time, and the radial acceleration represents the second-order rate of change of the target slant range at the motion parameter reference time.

[0070] In this embodiment, each candidate motion parameter is used to generate the predicted distance deviation of the target relative to the motion parameter reference time. The processor converts the predicted distance deviation into the displacement of the distance sampling points and uses 8-point windowed sine interpolation to obtain the corrected complex sampled values ​​from the local range processing complex echo. When the distance sampling points required for interpolation exceed the effective distance range, the corresponding output sampling points are marked as invalid and do not participate in coherent accumulation with zero values. The stationary scattering unit performs the same distance sampling position correction according to the stationary distance trajectory.

[0071] In this embodiment, the candidate motion parameters are selected using a coarse-to-fine finite value method. The range of the coarse candidate radial velocity is... 15 m / s to 15 m / s, in intervals of 0.5 m / s; the coarse candidate radial acceleration range is The range is 2 m / s² to 2 m / s², with an interval of 0.2 m / s². The processor first performs distance migration correction and focusing on the coarse candidate parameters, and then sets up a fine candidate range around the coarse candidate parameters with better quality items within the region; the fine candidate radial velocity interval is 0.1 m / s, and the fine candidate radial acceleration interval is 0.05 m / s², with each fine candidate range covering one coarse interval on both sides of the coarse candidate value. The fine candidate interval is determined based on the maximum residual distance deviation caused by half a fine candidate interval not exceeding 0.5 distance sampling points.

[0072] In this embodiment, the stationary branch performs focusing pixel-by-pixel according to a unified imaging coordinate grid. The processor calculates the corresponding stationary slant range based on the radar antenna phase center position and the position of the pixel to be imaged for each pulse. The stationary slant range is converted into the range sampling position in the local range processing complex echo, obtaining the complex sampled value after range migration correction. The propagation phase corresponding to the stationary slant range is subtracted, and then the result is accumulated in a complex manner according to the validity weight in the target-level phase-connected pulse index set to obtain the stationary parameter version complex value of the pixel. After performing the same processing on all pixels in the region, a stationary parameter version complex image is obtained.

[0073] In this embodiment, the motion branch incorporates the distance changes determined by candidate radial velocities and candidate radial accelerations into the stationary slant range corresponding to the pixel to be imaged. Local distance processing complex echoes are obtained according to the corrected distance position. The propagation phase corresponding to the candidate motion distance trajectory is subtracted, and then complex summation is performed according to the validity weights in the target-level phase-connected pulse index set. Each set of candidate radial velocities and candidate radial accelerations generates a motion parameter version of the complex image. Different parameter versions of the complex images use the same target connected region boundary, output pixel spacing, coordinate origin, and coordinate axis direction.

[0074] In some embodiments, the range pixel spacing and azimuth pixel spacing of the unified imaging coordinate grid are each no greater than half of the corresponding theoretical resolution. In this embodiment, both the range pixel spacing and azimuth pixel spacing are set to 0.25m. The output pixel spacing is pre-stored according to the radar operating mode and recorded with multiple images of different parameter versions, without changing between different parameter versions of the same target connected region.

[0075] In this embodiment, the stationary scattering unit retains the stationary branch parameter version complex image and the motion branch verification version with radial velocity and radial acceleration of 0; the motion scattering unit retains the stationary branch reference version and the motion branch parameter version corresponding to each candidate motion parameter. The stationary branch reference version is used to compare the residual distance migration and azimuth main lobe width before and after motion compensation, and will not replace the qualified motion branch parameter version if the imaging quality fails to meet the qualification conditions.

[0076] In this embodiment, the parameter dependency graph is stored using a directed acyclic structure. Each parameter node and result node represents a specific processing instance. Parameter nodes record the parameter name, parameter value, working mode identifier, local distance processing output interval identifier or target connected region identifier, and parameter version identifier; result nodes record the input data identifier, output data identifier, processing object range, and result version identifier. Parameters with the same name but acting on different local distance processing output intervals or different target connected regions are established as separate nodes.

[0077] In this embodiment, the parameter nodes include a fixed reference range window parameter node, a local range processing output interval parameter node, a first range window parameter node, a second range window parameter node, a group delay correction parameter node, a phase correction parameter node, a dynamic fusion weight parameter node, a phase continuity determination parameter node, a segment connectivity determination parameter node, a target-level phase connectivity pulse index set node, a range migration correction parameter node, and a branch focusing parameter node. The result nodes include a pre-detection complex range image node, two branch result nodes, a local range processing complex echo node, a target connectivity region node, a phase continuity segment node, a phase connectivity aperture map node, a range migration correction result node, a parameter version complex image node, and a quality verification result node.

[0078] In this embodiment, when a processing result is generated by one or more parameters or preceding processing results, directed connections are established from the corresponding parameter node or preceding result node to the processing result node. The first distance window parameter node points to the high-resolution distance branch result node, the second distance window parameter node points to the sidelobe suppression branch result node, and the two branch result nodes, the group delay correction parameter node, the phase correction parameter node, and the dynamic fusion weight parameter node all point to the local distance processing complex echo node. The target-level phase connectivity pulse index set node and the distance migration correction parameter node point to the distance migration correction result node, and the distance migration correction result node and the branch focusing parameter node point to the parameter version complex image node.

[0079] like Figures 6-8 As shown, in this embodiment, when parameters are adjusted, the original parameter nodes and result nodes are not overwritten. Instead, new versions of parameter nodes and result nodes are generated, while retaining the substitution relationship between the old and new versions. The quality verification and image output module can backward search for the parameter nodes it depends on from the complex image of the parameter version that has not passed verification, and it can also forward search for all downstream result nodes from the adjusted parameter nodes. When adjusting the candidate radial velocity of the target connected region of the traveling vehicle, the nodes to be recalculated only include the distance migration correction result, the complex image of the motion branch parameter version, and the quality verification result of the target connected region of the traveling vehicle, and do not include the processing results of other target connected regions.

[0080] In this embodiment, the acceptable imaging quality conditions within the region include range main lobe width, azimuth main lobe width, peak sidelobe ratio, integral sidelobe ratio, and residual range migration. The quality verification and image output module preferentially selects point scattering units that meet the isolation conditions from the complex image of the parameter version. The isolation condition is that the peak amplitude of the scattering unit reaches 10 times the local noise amplitude, and the interval between it and other local peaks in both the range and azimuth directions is not less than 3 times the corresponding main lobe width. The 10 times and 3 times are determined by joint calibration of a stationary corner reflector and adjacent weak scatterers to reduce the influence of adjacent scattering responses on the main lobe and sidelobe measurements.

[0081] In this embodiment, both the distance main lobe width and the azimuth main lobe width are measured on the power map. The boundary is defined by the position where the power on both sides of the main peak drops to half the power of the main peak, and linear interpolation is performed between adjacent pixels. The peak sidelobe ratio is the decibel value of the ratio of the highest sidelobe amplitude outside the main lobe range to the main peak amplitude. The integral sidelobe ratio is the decibel value of the ratio of the total energy outside the main lobe range to the total energy within the main lobe range. The main lobe range is defined by the first local minima on both sides of the main peak.

[0082] In this embodiment, the remaining distance migration is determined based on the actual distance trajectory after distance migration correction. The processor calculates the deviation of the remaining distance position of each valid pulse from its average distance position, and takes the maximum absolute deviation as the remaining distance migration. The remaining distance migration limit is set at 0.5 distance sampling points, and its determination condition is that the main lobe widening caused by the remaining distance migration does not exceed the repeated measurement fluctuation range.

[0083] In some embodiments, when no point scattering units satisfying the isolation conditions exist within the target connected region, the range main lobe width, azimuth main lobe width, peak sidelobe ratio, and integral sidelobe ratio are verified using the calibration point response with the same range window, effective pulse weight, range migration interpolation method, and branch focusing parameters as the current parameter version, and the source of the quality data is marked in the quality record. The remaining range migration of the current target connected region is still calculated based on the actual range trajectory of that region. This processing is applicable to extended scattering regions corresponding to road edges and building outlines.

[0084] In this embodiment, the range main lobe width limit and the azimuth main lobe width limit are determined based on the theoretical range resolution and the theoretical azimuth resolution corresponding to the processing Doppler bandwidth of the current radar operating mode, respectively, and are both taken as 1.2 times the corresponding theoretical values. The peak sidelobe ratio limit is set to no higher than... 22dB, the integral sidelobe ratio limit is set to no higher than 100dB. 12dB. The limiting value is determined by the joint calibration results of the strong scattering angle reflector and the adjacent weak scattering angle reflector, with the main peak of the weak scattering angle reflector being able to separate from the side lobes of the strong scattering angle reflector as the determining condition.

[0085] In this embodiment, the quality criteria for inter-region stitching include the degree of complex correlation in the overlapping region and the absolute value of the remaining phase error in the overlapping region. The processor only uses common effective pixels whose amplitude reaches three times the local noise amplitude in the overlapping portion of the output results of the two candidate regions for calculation. The degree of complex correlation in the overlapping region is determined by complex echo multiplication accumulation and energy normalization. The processor obtains the phase of the complex echo multiplication accumulation result of the output results of the two candidate regions, subtracts the already applied regional phase alignment amount, and adjusts the obtained phase to... Within a range of 180° to 180°, the absolute value is taken as the absolute value of the remaining phase error in the overlapping area. When there are fewer than 64 common effective pixels, the overlapping part is marked as lacking stitching verification data, complex fusion is not performed, and only the output results of the area with better quality items within the region are retained.

[0086] In this embodiment, the complex correlation degree limit for the overlapping region is set to be no less than 0.85, and the absolute value limit for the remaining phase error in the overlapping region is set to be no more than 45°. These two limits are determined using corner reflector data with overlapping imaging ranges. After complex stitching under these limits, the amplitude drop of the main peak in the overlapping region does not exceed 0.5 dB. If any item in the imaging quality qualification conditions within a region fails, the current parameter version of the complex image is not output as a candidate region; if any item in the inter-region stitching quality qualification conditions fails, the current stitching result is not retained.

[0087] In this embodiment, the distance window and dynamic fusion weight processing parameter set includes a first distance window parameter, a second distance window parameter, a protected area width, a dynamic fusion weight, and a weight transition area width; the phase consistency processing parameter set includes a phase continuity determination threshold, a phase change determination threshold, a segment connectivity determination threshold, a segment phase error threshold, a target-level phase connectivity pulse index set, and a region phase alignment amount; the distance migration correction processing parameter set includes a stationary distance trajectory, candidate radial velocity, candidate radial acceleration, and distance interpolation parameters; and the branch focusing processing parameter set includes an imaging coordinate grid, an effective aperture range, a propagation phase compensation amount, and a complex accumulation weight.

[0088] In this embodiment, a deterministic priority mapping relationship is adopted between quality items and processing parameter groups. When the range main lobe width, peak sidelobe ratio, or integral sidelobe ratio is unqualified, the range window and dynamic fusion weight processing parameter group are adjusted first; when the average complex correlation degree, the absolute value of the maximum adjacent pulse remaining phase difference, the complex correlation degree of the overlapping region, or the absolute value of the remaining phase error of the overlapping region is unqualified, the phase consistency processing parameter group is adjusted first; when the remaining range migration is unqualified, the range migration correction processing parameter group is adjusted first; when the azimuth main lobe width is unqualified but the remaining range migration is qualified, the branch focusing processing parameter group is adjusted first.

[0089] In this embodiment, when multiple quality items in a complex image with the same parameter version fail the verification, the processing parameter group is selected and adjusted in the upstream-to-downstream order of the parameter dependency graph, namely the distance window and dynamic fusion weight processing parameter group, the phase consistency processing parameter group, the distance migration correction processing parameter group, and the branch focus processing parameter group. After completing one incremental recalculation, the imaging quality item verification for the entire region is re-executed. Only after the quality item corresponding to the current processing parameter group passes the verification is the next quality item that fails the verification processed.

[0090] In this embodiment, after adjusting a set of processing parameters, the quality verification and image output module searches forward along the directed connection edges from the corresponding parameter nodes, marking the reachable result nodes as nodes to be recalculated. When the distance window and dynamic fusion weight processing parameter sets are adjusted, the two branch results of the affected local distance processing output interval, the local distance processing echo, and the target connected region dependent on the echo are recalculated. When the phase consistency processing parameter set is adjusted, the recalculation starts from the phase continuity segment or regional phase alignment node of the affected target connected region. When the distance migration correction processing parameter set or the branch focusing processing parameter set is adjusted, the recalculation scope is limited to the corresponding target connected region and the corresponding parameter version.

[0091] In this embodiment, the maximum number of adjustments for each processing parameter group is 3. This number is determined by the changes in quality items with the number of adjustments in 20 sets of calibration data. After the 3rd adjustment, the quality improvement obtained by further adjustments does not exceed the repeated measurement fluctuation range, so the number of adjustments is not increased. If no qualified parameter version is found after 3 adjustments, the corresponding target connected region is marked as having no valid region in the output result, and the complex image of the parameter version that failed the verification is not used as a replacement.

[0092] In this embodiment, complex images of parameter versions that pass the imaging quality qualification criteria for all regions are selected from each group of complex images of region parameter versions. If only one qualified parameter version exists, it is determined as the candidate region output result. If multiple qualified parameter versions exist, the azimuth main lobe width, range main lobe width, peak sidelobe ratio, integral sidelobe ratio, and residual range migration are compared sequentially. If the difference of the current comparison item does not exceed the corresponding repeated measurement fluctuation range, the next item is compared.

[0093] In this embodiment, the range of repeated measurement fluctuations is determined by processing the same corner reflector data 20 times. The maximum repeated measurement fluctuations for the range main lobe width, azimuth main lobe width, peak sidelobe ratio, integral sidelobe ratio, complex correlation degree in the overlapping region, and absolute value of the residual phase error in the overlapping region are 0.02m, 0.03m, 0.3dB, 0.3dB, 0.02, and 3°, respectively. When the difference in quality items between multiple qualified parameter versions does not exceed the corresponding value, the item is considered the same, and the comparison continues to the next quality item; when all items are considered the same, the parameter version with the earlier generation time is selected.

[0094] In this embodiment, the quality verification and image output module writes the candidate region output results into a unified imaging coordinate system in a fixed order, from smallest to largest, according to the target connected region identifiers. When the candidate region output results to be written overlap with the already written region output results, the average complex phase difference is calculated using common effective pixels whose amplitude reaches three times the local noise amplitude, and a complex phase rotation in opposite directions is applied to the candidate region output results to be written. Subsequently, linear stitching weights are determined according to the distance from the pixel to the boundaries of the two regions, so that the sum of the two stitching weights at the same pixel is 1, and complex domain stitching is performed.

[0095] In this embodiment, the complex correlation degree of the overlapping region and the absolute value of the remaining phase error of the overlapping region are recalculated after stitching. When both quality items meet the inter-region stitching quality qualification conditions, the stitching result is retained; if either item fails, other parameter versions that have passed the regional verification within the current target connected region are tried in turn. If no qualified stitching result is found, the valid pixels of the region written first are retained, the overlapping part of the current target connected region is marked as invalid, and the item and parameter version identifier that failed the verification are saved in the quality record.

[0096] In a specific application scenario, an airborne synthetic aperture radar (SAR) is used to perform high-resolution imaging of a ground observation area including warehouse buildings, metal facilities, roads, and vehicles. The corners of warehouse buildings and metal facilities form strong scattering elements, road edges and small ground features form weak scattering elements, and vehicles form moving scattering elements with varying range and phase histories. The radar carrier frequency is 9.6 GHz, the transmitted linear frequency modulated signal bandwidth is 300 MHz, the pulse width is 20 μs, the range sampling frequency is 360 MHz, the pulse repetition frequency is 1200 Hz, the radar platform flight speed is 120 m / s, the center slant range of the observation area is 10000 m, and 2400 pulses are continuously acquired.

[0097] In this embodiment, based on a 300MHz signal bandwidth, the theoretical range resolution is approximately 0.5m; based on a 360MHz range sampling frequency, the slant range interval between adjacent range sampling points is approximately 0.4167m; 2400 pulses correspond to a 2s observation duration, during which the radar platform flies 240m. Based on the processing Doppler bandwidth and observation geometry of the current operating mode, the theoretical azimuth resolution is calibrated to 0.65m; therefore, the range mainlobe width is limited to 0.60m, and the azimuth mainlobe width is limited to 0.78m.

[0098] In this embodiment, each local range processing output interval contains 64 range sampling points, and adjacent local range processing output intervals overlap by 32 range sampling points. Each local range processing output interval corresponds to a slant range length of approximately 26.667m, and the overlapping portion corresponds to a slant range length of approximately 13.333m. The 20μs reference signal contains 7200 sampling points at a sampling frequency of 360MHz. Therefore, when performing range matched filtering, a range processing block of no less than 8192 points is used. The length of the processing block is not less than the sum of the effective response length of the range matched filter and the required output interval length minus 1, and the overlapping retention method is used to obtain the corresponding 64 output sampling points.

[0099] In this embodiment, the working mode parameters were calibrated using 30 sets of echoes without obvious targets, 30 sets of echoes from stationary corner reflectors, 30 sets of combined strong and weak corner reflector echoes, and 20 sets of echoes from controlled moving vehicles. The noise region determination threshold was set to a normalized amplitude of 0.018, the strong scattering region determination threshold was set to a normalized amplitude of 0.42, and the protection zone width was determined by taking 12 distance sampling points on each side of the strong scattering region, corresponding to a single-sided slant distance of approximately 5.000m. The Kaiser window shape parameter for the first distance window was set to 0, and the Kaiser window shape parameter for the second distance window was set to 3.

[0100] In this embodiment, after the two branches complete group delay correction and phase correction, the distance to the main peak position difference is 0.04 distance sampling points, and the complex phase difference of the main peak is 1.6°, which respectively meet the requirements of not exceeding 0.1 distance sampling points and not exceeding 5°. The dynamic fusion weight of the sidelobe suppression branch in the strong scattering region and the protection region is 0.75, and the dynamic fusion weight of the sidelobe suppression branch in the noise region and the weak scattering region far from the strong scattering region is 0. A linear weight transition zone of 8 distance sampling points is set on both sides of the region boundary. The adjacent local distance processing output intervals use the overlapping part of 32 distance sampling points for complex phase alignment and linear complex splicing to obtain the local distance processing complex echo within the complete distance range.

[0101] In this embodiment, the phase connectivity aperture determination module uses three times the local noise amplitude as the effective detection threshold, and merges distance sampling points whose amplitude reaches the effective detection threshold and are continuous in both the distance and slow-time directions into target connected regions. Warehouse corners, road edges, and moving vehicles each form corresponding target connected regions. When the complex correlation between adjacent pulses reaches 0.85 and the absolute value of the remaining phase difference between adjacent pulses does not exceed 45°, adjacent pulses are classified into the same continuous phase segment.

[0102] In this embodiment, the phase connectivity aperture determination module sets each phase continuous segment as a node in the phase connectivity aperture graph. When the number of interrupt pulses between two phase continuous segments belonging to the same target connectivity region does not exceed 4, the complex correlation between segments reaches 0.85, and the absolute value of the remaining phase difference between segments does not exceed 45°, a connection edge is established between the corresponding nodes. The processor deletes the set of connected nodes with fewer than 480 valid pulses or an aperture time span of less than 1.67s.

[0103] In this embodiment, 1864 valid pulses were obtained from the target connectivity region at the corner of the warehouse building, with the first and last valid pulses covering an aperture time span of 1.85 s; 1728 valid pulses were obtained from the target connectivity region at the road edge, with the first and last valid pulses covering an aperture time span of 1.75 s; and 936 valid pulses were obtained from the target connectivity region at the traveling vehicle, with the first and last valid pulses covering an aperture time span of 1.80 s. All three target connectivity regions satisfy the conditions of having at least 480 valid pulses and an aperture time span of at least 1.67 s. The target-level phase connectivity pulse index set stores the pulse number and corresponding validity weight according to the pulse acquisition time.

[0104] In this embodiment, the maximum distance deviation between the actual distance trajectory and the stationary distance trajectory of the corner of the warehouse building and the edge of the road does not exceed one distance sampling point, and the absolute value of the remaining phase error in the stationary state does not exceed 45°. Therefore, they are identified as stationary scattering units. The maximum distance deviation of the connected area of ​​the moving vehicle target is 18.6 distance sampling points, corresponding to approximately 7.75m, which exceeds the stationary state determination threshold of one distance sampling point. Therefore, it is identified as a moving scattering unit.

[0105] In this embodiment, the processor performs minimum residual fitting on the distance difference between the actual distance trajectory of the moving vehicle and the stationary distance trajectory, obtaining an initial radial velocity of 8.47 m / s and an initial radial acceleration of 0.38 m / s². After coarse candidate parameter processing, the parameter versions corresponding to 8.5 m / s and 0.4 m / s² have smaller residual distance migration and azimuth main lobe width. The processor then performs fine candidate processing around these coarse candidate parameters at intervals of 0.1 m / s and 0.05 m / s², ultimately determining the parameter versions with a radial velocity of 8.5 m / s and a radial acceleration of 0.4 m / s².

[0106] In this embodiment, in the static parameter version complex image corresponding to the corner point of the warehouse building, the distance main lobe width is 0.54m, the azimuth main lobe width is 0.62m, and the peak-to-side-lobe ratio is... 23.1 dB, integral sidelobe ratio is The image quality was 13.4 dB, and the remaining distance migration was 0.18 distance sampling points. All of these points met the imaging quality qualification conditions for the corresponding area, so they were determined as candidate areas for the corner points of the warehouse building.

[0107] In this embodiment, the static parameter version complex image corresponding to the road edge uses a calibration point response measurement of the main lobe and side lobes with the same distance window, effective pulse weight, and focusing parameters as the current parameter version. The distance main lobe width is 0.57m, the azimuth main lobe width is 0.66m, and the peak-to-side-lobe ratio is [missing value]. 22.8dB, integral sidelobe ratio is 12.9dB; The remaining distance migration corresponding to the actual distance trajectory at the road edge is 0.21 distance sampling points, all of which meet the imaging quality qualification conditions in the corresponding area. Therefore, it is determined as the candidate area output result for the road edge.

[0108] In this embodiment, the vehicle exhibits residual distance migration at 18.6 distance sampling points in the stationary branch reference version, with an azimuth main lobe width of 1.46m. Therefore, it fails the regional imaging quality check. The quality check and image output module maps the residual distance migration to the distance migration correction processing parameter group and, based on the parameter dependency graph, recalculates only the motion distance migration correction result, the motion branch parameter version complex image, and the quality check result for the connected region of the vehicle target.

[0109] In this embodiment, in the complex image of motion parameters corresponding to a radial velocity of 8.5 m / s and a radial acceleration of 0.4 m / s², the distance main lobe width is 0.56 m, the azimuth main lobe width is 0.71 m, and the peak-to-side-lobe ratio is [missing value]. 22.6dB, integral sidelobe ratio is With a resolution of 12.5dB and a remaining distance migration of 0.32 distance sampling points, all of which meet the imaging quality requirements within their respective areas, this parameter version of the complex image is determined as the candidate area output result for the moving vehicle.

[0110] In this embodiment, the quality verification and image output module writes the candidate region output results corresponding to the corners of warehouse buildings, road edges, and driving vehicles into a unified imaging coordinate system according to the target connected region identification order. When the output results of adjacent candidate regions overlap, the average complex phase difference is determined using common effective pixels and complex phase alignment is performed. Then, complex domain stitching is performed using stitching weights that vary linearly with the distance to the region boundaries and sum to 1. After stitching, the complex correlation degrees of each overlapping region are 0.91, 0.89, and 0.88, respectively, and the absolute values ​​of the remaining phase errors of the corresponding overlapping regions are 18°, 24°, and 31°, respectively. All of these meet the requirements that the complex correlation degree of the overlapping regions is not less than 0.85 and the absolute value of the remaining phase error of the overlapping regions does not exceed 45°. Therefore, the stitching results are retained and the final high-resolution complex image is generated.

[0111] In some embodiments, invalid pixel markers are written to the locations of the no-valid-region output results in the final high-resolution complex image, and the parameter version complex image that has not passed the imaging quality qualification condition check is not used for filling. The final high-resolution complex image is associated with and saves the original complex echo data identifier, local distance processing parameters, target-level phase connectivity pulse index set, range migration correction parameters, branch type, parameter version identifier, and quality check result corresponding to each target connected region.

[0112] In some embodiments, the echo local distance processing module, the phase connectivity aperture determination module, the branch focusing and version generation module, and the quality verification and image output module are executed sequentially by the same processor, or they can be executed by multiple processors according to data dependencies. The echo local distance processing module outputs the local distance processed complex echo, region type, and dynamic fusion weights to the phase connectivity aperture determination module; the phase connectivity aperture determination module outputs the target connectivity region and the target-level phase connectivity pulse index set to the branch focusing and version generation module; the branch focusing and version generation module outputs the region parameter version complex image group and parameter dependency graph to the quality verification and image output module; the quality verification and image output module sends parameter adjustment instructions to the corresponding modules and receives incremental recalculation results.

[0113] In some embodiments, the system includes a processor, a radar data interface, a navigation data interface, a memory, and an image output interface. The radar data interface receives raw complex echo data and radar control parameters; the navigation data interface receives platform navigation parameters corresponding to the acquisition time of each pulse; and the memory stores range window parameters, threshold calibration results, group delay correction parameters, phase correction parameters, intermediate processing results, parameter dependency graphs, and the final high-resolution complex image. After reading the computer program from the memory, the processor sequentially executes the processing logic of the echo local range processing module, the phase connectivity aperture determination module, the branch focusing and version generation module, and the quality verification and image output module. The image output interface outputs the final high-resolution complex image and its corresponding quality record.

[0114] In some embodiments, the determination thresholds and processing parameters are obtained separately according to the radar operating mode during the radar calibration phase. The radar operating mode is determined by the carrier frequency, signal bandwidth, pulse width, range sampling frequency, pulse repetition frequency, platform velocity range, and observation slant range range. Each radar operating mode collects echoes from sources including no obvious target echoes, stationary corner reflector echoes, combined strong and weak corner reflector echoes, and controlled moving vehicle echoes. Echoes from sources without obvious targets are used to determine the noise region determination threshold and effective detection threshold; echoes from stationary corner reflectors are used to determine local range processing output interval parameters, guard zone width, group delay correction allowable error, phase correction allowable error, phase continuity determination threshold, minimum coherent accumulation pulse number, and minimum aperture time span; combined strong and weak corner reflector echoes are used to determine the first range window parameters, second range window parameters, dynamic fusion weight, peak sidelobe ratio limit, and integral sidelobe ratio limit; echoes from controlled moving vehicles are used to determine the stationary state determination threshold, candidate radial velocity range, candidate radial acceleration range, and candidate parameter interval. The calibration results are saved with the radar operating mode identifier as an index. During actual imaging, the corresponding calibration results are retrieved based on the radar operating mode identifier carried by the original complex echo data.

[0115] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A radar high-resolution imaging method based on dynamic windowing processing, characterized in that, include: The original complex echo data is acquired, and a pre-detection complex range image is generated based on the original complex echo data. The noise region, strong scattering region, weak scattering region and guard region are determined based on the pre-detection complex range image. Range high-resolution branch processing and sidelobe suppression branch processing are performed on the original complex echo data. After group delay correction, phase correction and dynamic fusion, the locally range-processed complex echo is obtained. Based on the local distance processing echo, the target connected region and phase continuity segment are determined, a phase connectivity aperture map is established, and the target-level phase connectivity pulse index set corresponding to each target connected region is extracted from the phase connectivity aperture map; Based on the target-level phase connectivity pulse index set, range migration correction and stationary or moving branch focusing are performed on the local range processing complex echo to generate a region parameter version complex image group, and a parameter dependency graph between each processing parameter and the region parameter version complex image group is established. The complex image group of the region parameter version is checked according to the imaging quality qualification conditions. The quality items that fail the check are mapped to the corresponding processing parameter group. The affected processing nodes are incrementally recalculated according to the parameter dependency graph. The region output result is determined from the parameter version complex image that passes the check. The final high-resolution complex image is generated according to the region output result.

2. The radar high-resolution imaging method based on dynamic windowing processing according to claim 1, characterized in that, Determining the noise region, the strong scattering region, the weak scattering region, and the protection region includes: A fixed reference distance window is used to perform pre-detection range compression on the original complex echo data to obtain the pre-detection complex range image, and the pre-detection complex range image is divided into overlapping local range processing output intervals along the range direction; The noise region determination threshold is determined based on the amplitude statistics of the targetless distance interval, and the strong scattering region determination threshold above the noise region determination threshold is determined based on the distance compression of the main peak amplitude of the calibrated strong scattering unit. Distance sampling points with amplitudes not lower than the strong scattering region determination threshold are defined as the strong scattering region. Distance sampling points located within the range of the distance sidelobes on both sides of the strong scattering region but not belonging to the strong scattering region are defined as the protection region. Among the remaining distance sampling points, distance sampling points with amplitudes not higher than the noise region determination threshold are defined as the noise region. The remaining distance sampling points are defined as the weak scattering region.

3. The radar high-resolution imaging method based on dynamic windowing processing according to claim 2, characterized in that, Obtaining the local range processed complex echo includes: For each of the local distance processing output intervals, an extended input interval covering the effective response length of the distance matched filter is extracted from the original complex echo data. Two distance matched filters are constructed using a first distance window and a second distance window with a higher degree of apodization than the first distance window. High-resolution distance branch processing and sidelobe suppression branch processing are performed on the extended input intervals. Output sampling points affected by convolution boundaries are deleted, and the complex processing results of the two branches corresponding to the local distance processing output intervals are retained. Based on the distance to the main peak position difference and the complex phase difference of the main peak after the calibration scattering unit is processed by two branches, group delay correction and phase correction are performed on the complex processing results of the two branches. Dynamic fusion weights are determined based on the regions to which each distance sampling point belongs. The complex processing results of the two branches after correction are fused point by point, and the overlapping parts of adjacent local distance processing output intervals are spliced ​​in the complex domain.

4. The radar high-resolution imaging method based on dynamic windowing processing according to claim 1, characterized in that, Determining the target connected region and the phase continuum segment includes: The effective detection threshold is determined based on the local noise amplitude of the local distance processed echo, and the distance sampling points whose amplitude reaches the effective detection threshold and are continuous in the distance direction and slow time direction are merged into the target connected region. In each target connected region, the effective distance sampling points in adjacent pulses are aligned with the distance trajectory, the degree of complex correlation between adjacent pulses is calculated, and the predicted phase difference corresponding to the platform motion geometry and the actual distance trajectory is subtracted from the observed phase difference to obtain the absolute value of the remaining phase difference between adjacent pulses. Continuous pulses whose complex correlation reaches the phase continuity determination threshold and whose absolute value of the remaining phase difference between adjacent pulses does not exceed the phase change determination threshold are classified into the same phase continuity segment.

5. The radar high-resolution imaging method based on dynamic windowing processing according to claim 4, characterized in that, Establishing the phase connectivity aperture map includes: Each of the phase continuity segments is set as a node in the phase connectivity aperture diagram; When the pulse ranges of two phase continuous segments belonging to the same target connected region are adjacent, or the number of interrupted pulses between the two pulse ranges does not exceed the allowed number of interruptions, the comparison pulses at the boundary of the two phase continuous segments are extracted respectively, the effective distance sampling points in the comparison pulses are aligned with the distance trajectory, and the degree of complex correlation between segments and the absolute value of the remaining phase difference between segments are determined based on the aligned complex echo. When the complex correlation between the segments reaches the segment connectivity determination threshold and the absolute value of the remaining phase difference between the segments does not exceed the segment phase error threshold, a connection edge is established between the corresponding nodes.

6. The radar high-resolution imaging method based on dynamic windowing processing according to claim 5, characterized in that, Extracting the target-level phase-connected pulse index set includes: In the phase connectivity aperture diagram corresponding to each target connectivity region, find the set of connected nodes and delete the set of connected nodes whose effective pulse count does not reach the minimum coherent accumulation pulse count or whose aperture time span does not reach the minimum aperture time span. The target set of connected nodes is determined according to the aperture time span, number of effective pulses, average complex correlation degree, absolute value of the remaining phase difference between the largest adjacent pulses, and number of pulse interruptions of the other connected node sets, in a preset comparison order. The pulse numbers covered by the target connected node set are arranged according to the acquisition time, and a validity weight is recorded for each pulse number to obtain the target-level phase connected pulse index set.

7. The radar high-resolution imaging method based on dynamic windowing processing according to claim 6, characterized in that, Generating the complex image group of the region parameter version and establishing the parameter dependency graph includes: Based on the maximum distance deviation between the actual distance trajectory and the stationary distance trajectory of the target connected region and the remaining phase error after deducting the stationary predicted phase, the target connected region is divided into stationary scattering units or moving scattering units, and distance migration correction is performed according to the corresponding target-level phase connected pulse index set. Static branch focusing is performed on the static scattering unit, and motion branch focusing is performed on the moving scattering unit according to a finite set of candidate motion parameters to obtain multiple parameter versions of complex images of the same target connected region; Record the local distance processing parameters, target-level phase connectivity pulse index set, distance migration correction parameters, and branch focusing parameters that each parameter version complex image depends on, and establish the parameter dependency graph according to the input-output relationship of the processing results.

8. The radar high-resolution imaging method based on dynamic windowing processing according to claim 7, characterized in that, Performing incremental recalculation and generating the final high-resolution complex image includes: The imaging quality within a region is qualified by using the distance main lobe width, azimuth main lobe width, peak sidelobe ratio, integral sidelobe ratio, and remaining range migration. The quality of stitching between regions is qualified by using the degree of complex correlation in the overlapping region and the absolute value of the remaining phase error in the overlapping region. Map the range main lobe width, peak sidelobe ratio, or integral sidelobe ratio that failed the verification to the range window and dynamic fusion weight processing parameter group; map the phase consistency quality items that failed the verification to the phase consistency processing parameter group; map the remaining range migration that failed the verification to the range migration correction processing parameter group; and map the azimuth main lobe width that failed the verification to the branch focusing processing parameter group. The downstream processing node pointed to by the corresponding processing parameter group is determined according to the parameter dependency graph. Only the downstream processing node is recalculated, and the output result of the region that passes the imaging quality qualification condition and the inter-region stitching quality qualification condition is written into the unified imaging coordinate system.

9. A high-resolution radar imaging system based on dynamic windowing processing, characterized in that, include: The echo local range processing module is used to acquire raw complex echo data, generate pre-detection complex range images, identify noise regions, strong scattering regions, weak scattering regions and protection regions, perform high-resolution range branch processing and sidelobe suppression branch processing, and obtain locally range processed complex echoes through group delay correction, phase correction and dynamic fusion. The phase connectivity aperture determination module is used to determine the target connectivity region and phase continuity segment based on the local distance processed complex echo, establish a phase connectivity aperture map, and extract the target-level phase connectivity pulse index set; The branch focusing and version generation module is used to perform distance migration correction and static or moving branch focusing based on the target-level phase connectivity pulse index set, generate a complex image group of region parameter versions and establish a parameter dependency graph. The quality verification and image output module is used to verify the complex image group of the region parameter version according to the imaging quality qualification conditions, map the quality items that fail the verification to the corresponding processing parameter group, perform incremental recalculation according to the parameter dependency graph, and generate the final high-resolution complex image based on the output results of the regions that pass the verification.

Images