Traffic radar-based high-capture-rate target detection method and device, and storage medium

Abstract

The invention discloses a high-capture-rate target detection method and device based on a traffic radar, and a storage medium, and relates to the technical field of radar data processing. The method comprises the following steps: dividing a detection matrix into a first region and a second region; performing beam forming and detection on the second area, detecting the first area, fusing detection results, and outputting a preliminary detection point cloud; generating a speed hypothesis based on the initial detection point cloud, and generating multiple groups of candidate signals; performing angle estimation processing on each group of candidate signals based on a difference virtual array principle; calculating the peak amplitude of each angle spectrum, and taking the angle value corresponding to the angle spectrum with the maximum peak amplitude as a correction angle; cross-frame matching and speed ambiguity resolution processing are carried out on fast and slow frame point clouds which are alternately generated by a radar according to a specific time sequence and have different beam types. The method solves the problems that in traffic radar detection, the far-field angle resolution is insufficient, information is inaccurate due to speed ambiguity, and point cloud matching fusion is difficult.

Description

Technical Field

[0001] This invention belongs to the field of radar data processing technology, and in particular relates to a high acquisition rate target detection method, device and storage medium based on traffic radar. Background Technology

[0002] With the rapid development of intelligent transportation systems, accurate, real-time, and reliable detection of road traffic flow has become a key technological support for realizing intelligent traffic management, road network planning, and optimization decisions. Millimeter-wave radar, especially high-frequency millimeter-wave radar operating near the 90GHz band, has become one of the core sensors for fixed-location traffic flow detection due to its advantages such as high resolution, minimal weather-related influences, and all-weather operation. Its core task is to continuously and stably output three-dimensional target point clouds containing accurate distance, speed, and angle information, providing a high-quality data source for subsequent target tracking, trajectory analysis, and traffic condition assessment.

[0003] In existing technologies, improving the detection performance of traffic radar, especially its ability to acquire targets at long distances with low reflectivity (i.e., "high acquisition rate"), mainly revolves around two technical approaches:

[0004] The first type of approach focuses on enhancing the signal detection level. For example, it involves optimizing the algorithm in the constant false alarm rate (CFAR) detection stage, or using digital beamforming (DBF) technology to superimpose the gain of the received channel signal to achieve incoherent accumulation of weak target energy, thereby increasing its detection probability in a single frame of data.

[0005] For example, patent document CN119414339A discloses a DBF enhancement detection method for weak targets in millimeter-wave traffic radar, which acquires more targets by setting up enhancement regions and performing multiple DBF scans. However, such methods mainly focus on acquiring more raw points or preliminary point clouds in the radar detection matrix. The target information (especially angle and velocity) output by these methods often has limited accuracy or is ambiguous, and is not an accurate three-dimensional parameter that can be directly used for subsequent processing. These preliminary point clouds, as intermediate results, cannot directly ensure the output of a continuous, stable, and parameter-accurate final detection point cloud in complex traffic scenarios.

[0006] The second type of approach expands performance boundaries by improving radar hardware waveforms and data fusion strategies. A common practice is to use an alternating wide-beam and narrow-beam transmission mode, aiming to balance wide field-of-view coverage with long-range, high-precision detection. However, this approach faces significant challenges in practical deployment: Firstly, the spatiotemporal alignment and accurate fusion algorithms for point cloud data are extremely complex due to the significant differences in beam shape, resolution, and effective detection range between wide-beam and narrow-beam systems. If the fusion algorithm takes too long to compute, it will fail to meet the stringent real-time requirements of traffic radar for high refresh rates (typically requiring output in the tens of milliseconds), leading to system performance bottlenecks. Secondly, the detection logic of existing radar systems is not sufficiently adaptable to road scenarios. Many systems have relatively fixed and idealized lane detection models (e.g., only supporting symmetrical rectangular areas), making it difficult to flexibly adapt to complex road geometries such as curves and channelized intersections, resulting in inaccurate detection area settings and a large number of invalid calculations or missed detections.

[0007] In summary, the existing technology suffers from the following systemic shortcomings that urgently need to be addressed:

[0008] (1) Disconnect between parameter accuracy and completeness: Existing weak target enhancement schemes mostly produce "intermediate results" in the detection stage, failing to systematically solve the core problems of velocity fuzz correction and high-precision angle estimation in the final output point cloud, resulting in insufficient accuracy and completeness of point cloud three-dimensional information (distance, velocity, angle).

[0009] (2) The contradiction between fusion performance and real-time performance: The fusion strategy of multi-source data such as wide and narrow beams is limited by the algorithm complexity, making it difficult to meet the high real-time output requirements of traffic radar while ensuring fusion accuracy.

[0010] (3) Lack of scene adaptability and adaptive capability: The detection logic lacks deep integration with the dynamic and configurable real road model, and cannot be adaptively adjusted according to the actual lane parameters. The level of intelligence is low, which limits the optimal detection performance under different road conditions.

[0011] Therefore, there is an urgent need for an innovative, system-level detection method that can not only effectively improve the detection sensitivity of weak targets, but also ensure that the final output point cloud has accurate and unambiguous three-dimensional parameters, and can adapt to complex road scenarios. At the same time, it can ensure a high refresh rate of data output with an efficient fusion mechanism, thereby truly realizing high acquisition rate and high reliability detection of traffic radar in complex environments. Summary of the Invention

[0012] In view of the above-mentioned defects in the existing technology, the purpose of this invention is to provide a high acquisition rate target detection method, device and storage medium based on traffic radar, so as to systematically solve the problem of how to adaptively improve the weak target detection capability without reducing the radar data refresh rate in complex road scenarios, and finally output a three-dimensional target point cloud with accurate, complete and continuous parameters.

[0013] This invention solves the above-mentioned technical problems through the following technical solution: a high acquisition rate target detection method based on traffic radar, comprising:

[0014] Based on road geometry parameters, the detection matrix generated from radar echo signals is divided into a first region and a second region;

[0015] Beamforming and detection based on road geometry parameters are performed on the data in the second region, while the signal in the first region is detected simultaneously;

[0016] By fusing the detection results from the second region and the first region, a preliminary detection point cloud containing target distance and velocity information is output.

[0017] Based on the velocity information of each target in the preliminary detection point cloud and the radar's maximum unambiguous velocity, at least one velocity hypothesis is generated, and based on the velocity hypothesis, multiple sets of candidate signals are generated through Doppler phase compensation.

[0018] Each candidate signal is subjected to high-resolution angle estimation processing based on the differential virtual array principle to obtain multiple angle spectra;

[0019] Calculate the peak amplitude of each angle spectrum, and use the angle value corresponding to the angle spectrum with the largest peak amplitude as the correction angle of the target, thereby obtaining a detection point cloud containing target distance, velocity information and correction angle;

[0020] Based on the target distance and correction angle, cross-frame matching and velocity deblurring are performed on the fast and slow frame point clouds generated by the radar in a specific time sequence with different beam types. After fusion, the final target point cloud with accurate parameters is output.

[0021] This invention introduces road geometry parameters as core input, transforming radar detection logic from a fixed program into a dynamically configurable intelligent strategy. Based on the actual lane width, direction, and length, the radar can precisely divide the detection matrix into a high-priority long-range area (the second area) requiring enhanced detection and a short-range area requiring conventional processing (the first area). This not only ensures the detection range closely matches the actual road contour, completely eliminating the constraints of symmetrical rectangles, but more importantly, it enables on-demand allocation of computing power: only computationally intensive adaptive beamforming is applied to the low signal-to-noise ratio, high-priority long-range weak target areas, while efficient conventional detection is used for the short-range areas of strong targets. Thus, while significantly improving the detection capability of long-range weak targets, it effectively avoids the huge computational redundancy caused by full-range beamforming, ensuring processing efficiency from the source and laying a solid foundation for meeting high refresh rate requirements.

[0022] This invention effectively suppresses high sidelobe interference caused by physically sparse arrays through high-resolution angle estimation processing based on the principle of differential virtual arrays, obtaining an initial angle estimate with a sharp main lobe and high resolution. It creatively designs a correction process of "generating multiple sets of candidate signals → parallel processing to obtain multiple angle spectra → comparing peak values ​​for decision-making." This mechanism fundamentally corrects the systematic angle error introduced by velocity ambiguity, ensuring that the final output corrected angle has both high accuracy and high reliability. By associating the distance-velocity information of the initial detection point cloud with the original array signal, a precise angle is calculated through closed-loop signal processing, thus outputting complete and reliable three-dimensional target information of "distance-velocity-angle".

[0023] Because the angle used for cross-frame matching is a corrected, high-precision value, the trial-and-error calculations and ambiguities in the matching process are greatly reduced, making the association of fast and slow frame point clouds fast and accurate, laying a solid foundation for subsequent speed deambiguation. The processing object of this invention is fast and slow frame point clouds generated alternately according to a specific time sequence, indicating that its underlying architecture employs efficient system architectures such as sliding frame buffering and pipeline scheduling. This makes it possible to stably and quickly fuse multi-beam data and output high-frame-rate point clouds with limited resources, thus resolving the fundamental contradiction that fusion time affects the system refresh rate.

[0024] In this invention, adaptive detection provides accurate "distance-velocity" guidance information for angle correction. High-precision angle correction is the fundamental prerequisite for efficient fusion to be completed quickly and accurately. Ultimately, it ensures that targets from near to far and from strong to weak can be stably perceived, accurately measured, and continuously tracked. In complex traffic scenarios, it continuously outputs high-quality target trajectories, thereby achieving a comprehensive improvement in the systematic and highly reliable capture rate.

[0025] Furthermore, based on road geometry parameters, the detection matrix generated from radar echo signals is divided into a first region and a second region, including:

[0026] Perform Fourier transforms on the radar echo signal in the range and velocity dimensions to generate a range-Doppler detection matrix;

[0027] Based on the lane length in the road geometry parameters, a distance division threshold is set;

[0028] The regions in the range-Doppler detection matrix whose distance dimension subscripts are less than or equal to the distance division threshold are classified as the first region, and the regions whose distance dimension subscripts are greater than the distance division threshold are classified as the second region.

[0029] This invention dynamically divides the detection matrix into a near-field (first region) and a far-field (second region) by setting a distance threshold, allowing for differentiated detection algorithms for the two regions. This avoids uniformly executing computationally complex algorithms (such as beamforming based on road geometry) across the entire detection matrix, thereby significantly reducing the overall computational load and improving the system's processing speed and real-time performance.

[0030] Employing more direct or refined detection methods in the first region (near field) better addresses the challenges of dense target distribution, high signal-to-noise ratio, and complex multipath and occlusion effects in the near field. This helps reduce missed detections and false alarms, improving the reliability of detecting key near-field targets. Resource-intensive road geometry-based beamforming and detection are then focused on the second region (far field). Far-field targets require high angular resolution and are subject to strong road geometry constraints. This division ensures high-precision, targeted angle estimation and detection of far-field targets while avoiding unnecessary complex calculations in the near field, achieving an optimal match between algorithm resources and scene requirements.

[0031] Furthermore, beamforming and detection based on road geometry parameters are performed on the data within the second region, including:

[0032] The angle scanning range is calculated based on the lane width and preset distance division threshold in the road geometry parameters.

[0033] Based on the angle scanning range, a set of discrete beam steering vectors are generated;

[0034] Using the beam steering vector, the multi-channel complex signal extracted from the second region is beam synthesized to obtain complex signals in multiple beam directions;

[0035] The signal power is obtained by performing a modulo-squaring operation on the complex signal in each beam direction;

[0036] The signal power is incoherently accumulated within the coherent processing interval to obtain the accumulated energy distribution;

[0037] The accumulated energy distribution is subjected to constant false alarm rate detection, and the target detection result of the second region is output.

[0038] This invention dynamically calculates the angular scanning range based on prior information about lane width, and generates discrete beam steering vectors within this range for focused beamforming. This allows limited signal processing energy to be concentrated and projected onto the spatial orientation where the target is most likely to appear. This significantly suppresses noise and interference from irrelevant directions, thereby achieving higher angular resolution and detection signal-to-noise ratio in the far-field region.

[0039] Compared to traditional full-space scanning, limiting the angular scanning range significantly reduces the number of beams that need to be generated and computed. Simultaneously, subsequent incoherent accumulation further smooths noise fluctuations and enhances target continuity. Combined with constant false alarm rate (CFAR) detection, this process greatly reduces the overall computational burden and false alarm probability of the system while maintaining detection sensitivity, achieving efficient and reliable far-field target detection.

[0040] This invention deeply integrates the crucial prior knowledge of road geometry into the signal processing chain. By directly converting parameters such as lane width and distance threshold into constraints for beamforming, the radar processing becomes highly adaptable to specific road scenarios, enhancing its adaptability to different road environments (such as different numbers of lanes and curves), and improving the practicality and reliability of the detection results.

[0041] Further, based on the stated angle scanning range, a set of discrete beam steering vectors is generated, including:

[0042] Within the angle scanning range, K discrete angle points are selected at preset intervals;

[0043] Calculate the beam steering vector at each discrete angle point based on the physical location of the radar receiving antenna array.

[0044] This invention allows for flexible adjustment of beamforming angular resolution and computational load by selecting discrete angle points at preset, controllable intervals within a defined angular scanning range. Smaller intervals enable more precise angular coverage, while reasonable interval settings can avoid generating redundant beams and significantly improve computational efficiency while ensuring lane coverage accuracy.

[0045] The steering vector for each angle is calculated based on the actual physical location of the radar receiving antenna array. This strictly adheres to the basic principles of array signal processing, ensuring the accuracy of beamforming in spatial pointing. Furthermore, the method of this invention is adaptable to receiving arrays of arbitrary configurations (such as uniform or sparse arrays), enhancing the algorithm's versatility and engineering practicality.

[0046] Furthermore, based on the velocity information of each target in the preliminary detection point cloud and the radar's maximum unambiguous velocity, at least one velocity hypothesis is generated, including: Obtain the velocity information of each target in the preliminary detection point cloud. and the maximum unambiguous speed of radar ; Based on the velocity information of each target in the preliminary detection point cloud and the maximum unambiguous speed of radar Generate at least one velocity assumption: If the speed information If the value is greater than 0, then the speed is assumed to be ; If the speed information If less than or equal to 0, then the speed is assumed to be ; And the velocity information of each target in the preliminary detection point cloud. This is directly used as the speed assumption.

[0047] This invention directly addresses the velocity folding phenomenon caused by the inherent maximum unambiguous velocity limit of radar. It utilizes a simple and deterministic mathematical rule (based on ±2 from the original velocity sign) to achieve this. The algorithm generates a most probable true velocity candidate value (i.e., velocity hypothesis) for each target that may experience velocity ambiguity (based on the offset of the velocity). This provides a clear and efficient entry point for solving the velocity ambiguity problem. This generation rule relies solely on the original velocity symbol, completely avoiding complex pattern recognition or iterative searches. Its logic is extremely simple and deterministic, free from ambiguity or nested branches, allowing it to execute stably and reliably under various signal-to-noise ratios and scenarios, demonstrating strong engineering robustness. The generated velocity hypothesis is not the final result, but rather provides the unique and necessary input for subsequent Doppler phase compensation. This step is a crucial bridge connecting "velocity detection" and "high-resolution angle estimation," ensuring that subsequent sophisticated algorithms such as differential virtual arrays can perform phase correction under the correct velocity hypothesis, thereby ultimately achieving accurate angle calculation.

[0048] Furthermore, based on the aforementioned velocity assumption, multiple sets of candidate signals are generated through Doppler phase compensation, including:

[0049] For each velocity assumption generated for each target, calculate its corresponding Doppler phase compensation value;

[0050] Phase compensation is performed on the original multi-channel received signal using the Doppler phase compensation values ​​described above.

[0051] Among them, the signals after phase compensation constitute the multiple candidate signals, and the original multi-channel received signal is the signal extracted from the detection matrix based on the distance and velocity information of the target.

[0052] In this embodiment, the core lies in transforming the abstract velocity assumption into a concrete "phase compensation value" that can be applied to the received signal using the Doppler effect formula. This allows the phase error between receiving channels caused by velocity ambiguity to be accurately corrected in the digital domain, laying a precise mathematical and physical foundation for subsequently recovering the correct spatial angle information using array signal processing technology.

[0053] Multiple phase compensation values ​​are calculated based on the velocity assumptions of the same target, and multiple sets of phase-corrected candidate signals are generated in parallel. This essentially pre-corrects the corresponding phase distortion for each possible velocity scenario at the signal level, so that the subsequent angle estimation module can objectively determine the velocity assumption that best matches the real situation based on a fair signal and comparison (such as peak amplitude), which greatly improves the accuracy and reliability of velocity defuzzification.

[0054] Furthermore, each candidate signal undergoes high-resolution angle estimation processing based on the differential virtual array principle, including:

[0055] For each group of candidate signals, calculate its corresponding spatial autocorrelation matrix;

[0056] The spatial autocorrelation matrix is ​​vectorized to obtain the initial differential signal vector;

[0057] The initial differential signal vector is subjected to redundancy removal and reordering of array element positions to obtain the corresponding equivalent uniform virtual array signal;

[0058] Spatial spectrum estimation is performed on the equivalent uniform virtual array signal to obtain an angular spectrum corresponding to the candidate signal.

[0059] In this embodiment, the core technology is differential virtual array technology. By calculating the autocorrelation matrix and quantizing it, an equivalent, longer-aperture continuous uniform virtual array can be reconstructed from the signal of a physically sparse array. This is equivalent to significantly expanding the effective aperture of the array without increasing the number of physical antennas, thereby fundamentally breaking through the angular resolution limit of the original array and achieving an ultra-high resolution angular estimation capability far superior to that of the physical array.

[0060] The "redundancy removal and reordering" step before generating the equivalent signal vector eliminates redundant and discontinuous array elements in the virtual array, ultimately resulting in a continuous and uniform virtual array. The beam pattern of the uniform array has a lower sidelobe level. Therefore, spatial spectrum estimation based on this equivalent signal vector can significantly suppress spurious sidelobes and grating lobes in the angular spectrum, making the main lobe of the real target sharper and more prominent, greatly improving the accuracy of angle estimation, single-target resolution, and reliability of multi-target detection.

[0061] The process of "autocorrelation → vectorization → virtual array reconstruction → spatial spectrum estimation" cleverly transforms the nonlinear optimization problem of super-resolution angle measurement based on sparse arrays into a linear problem of standard spatial spectrum estimation for an equivalent uniform linear array. This allows for the direct application of computationally efficient and stable classical spectrum estimation methods (such as FFT), avoiding the complex feature decomposition and search required by algorithms such as MUSIC and ESPRIT. While ensuring high performance, it significantly reduces the computational complexity and sensitivity to signal-to-noise ratio, making it more feasible for engineering implementation.

[0062] Furthermore, based on the target distance and correction angle, cross-frame matching and velocity de-ambiguity processing are performed on the fast and slow frame point clouds generated by the radar in a specific time sequence, including:

[0063] Maintain a buffer to cache target point cloud data of at least one historical subframe, the target point cloud data containing at least the target's distance index, correction angle, and velocity index;

[0064] Acquire the target point cloud data of the current subframe, and select the target point cloud data of the nearest historical subframe with the same beam type but different frame type from the buffer for matching;

[0065] Based on the distance subscript and the correction angle, the target point cloud data of the current subframe and the historical subframes are matched;

[0066] For successfully matched point cloud pairs, velocity deblurring is performed based on their velocity indices to determine the true velocity of the target.

[0067] In this embodiment, by maintaining a buffer and establishing explicit matching rules of "same beam, different frame type, nearest neighbor", the method of the present invention establishes a stable and reliable cross-frame association mechanism for point clouds of fast and slow frames generated continuously and asynchronously, ensuring that point cloud matching can achieve the best balance between temporal continuity and waveform differences, laying the foundation for subsequent joint calculation using different waveform characteristics.

[0068] The core objective of the matching rule in this invention is to pair observations of the same target across different waveforms (fast frames / slow frames). Since fast and slow frames have different velocity resolutions and ambiguity characteristics, this pairing directly provides two intrinsically related but differently expressed velocity observations of the same target, serving as a necessary and sufficient data prerequisite for accurate de-ambiguity processing using the velocity extension algorithm. The matching process considers not only target distance but also the precisely corrected angle processed by the aforementioned high-resolution algorithm, forming a two-dimensional (distance-angle) correlation constraint. Compared to one-dimensional matching using only distance, this invention significantly improves the uniqueness and accuracy of point cloud pairing, effectively overcoming matching errors caused by dense or closely spaced targets, and enhancing the system's robustness. The "nearest neighbor" condition in the matching rule ensures that the system always uses the most recent and most correlated historical frame data for matching, minimizing errors caused by data lag and effectively controlling the data size and processing latency of the buffer.

[0069] Furthermore, a weighted nearest neighbor algorithm is used to match the target point cloud data of the current subframe with those of historical subframes, including:

[0070] Calculate the normalized distance difference and normalized angle difference between the i-th point in the target point cloud data of the current subframe and the j-th point in the target point cloud data of the historical subframes, respectively.

[0071] The weighted fusion value is calculated based on the normalized distance difference and the normalized angle difference;

[0072] If the weighted fusion value is less than the preset matching threshold, then the i-th point and the j-th point are determined to be successfully matched, and the speed defuzzification process is initiated.

[0073] In this embodiment, the two key spatial parameters of distance and angle are combined through a weighted fusion method, overcoming the limitation of single-dimensional matching, which is prone to errors when targets are densely packed or trajectories intersect. Normalization processing ensures that parameters with different dimensions and different ranges of variation can be compared and fused fairly, enabling the matching criteria to adaptively take into account the proximity of targets in both distance and angle directions, significantly improving the comprehensiveness and accuracy of the matching.

[0074] By introducing a preset matching threshold, an objective and uniform threshold is set for successful matching. This effectively filters out accidental proximity points caused by noise, measurement errors, or unrelated targets, greatly reducing the false matching rate. The existence of weighting coefficients allows for adjusting the relative importance of distance and angle in matching according to the actual scenario (such as urban roads and highways), enhancing the algorithm's adaptability and robustness under different road conditions.

[0075] The matching method of this invention aims to filter point cloud pairs that are highly likely to originate from the same physical target for subsequent velocity deblurring steps. High-quality matching is a prerequisite for successful velocity deblurring. This invention ensures that only points with highly consistent spatial locations are deblurred through strict two-dimensional weighted threshold judgment. This fundamentally improves the data quality of the input for velocity deblurring processing, thereby guaranteeing the reliability and accuracy of the final velocity calculation result.

[0076] Furthermore, for successfully matched point cloud pairs, velocity deblurring is performed based on their velocity indices, including:

[0077] For a successfully matched point cloud pair, a velocity expansion matrix for the i-th point is generated based on the velocity index, the maximum velocity dimension index, and a preset expansion factor of the i-th point in the target point cloud data of the current subframe; a velocity expansion matrix for the j-th point is generated based on the velocity index, the maximum velocity dimension index, and a preset expansion factor of the target point cloud data of the historical subframe.

[0078] Calculate the absolute value of the difference between all pairs of elements in the velocity spread matrix of point i and the velocity spread matrix of point j, and select the minimum value among them;

[0079] If the minimum value is less than the preset speed threshold, the speed of the point cloud pair is determined to be successfully matched, and the speed corresponding to the speed index in the speed expansion matrix of the i-th point corresponding to the minimum value is taken as the true speed.

[0080] If the minimum value is greater than or equal to the preset velocity threshold, the velocity matching of the point cloud pair is determined to be unsuccessful, and it is retained until the velocity matching of the i-th point in the target point cloud data of the current subframe fails with all points in the target point cloud data of the historical subframe, at which point the i-th point is discarded.

[0081] In this embodiment, by generating a velocity expansion matrix containing the main lobe and multiple side lobes (velocity ambiguity terms), all possible values ​​of each velocity index within multiple velocity ambiguity periods are systematically enumerated. By calculating the minimum difference between all pairs of elements in the two expansion matrices, a unique pair of physically closest candidate true velocities can be accurately and automatically identified, thereby efficiently solving the ambiguity problem in radar velocity measurement and accurately restoring the true radial velocity of the target.

[0082] This invention introduces a "preset velocity threshold" as the final criterion. This not only effectively filters out false "minimum differences" caused by noise, matching errors, or unrelated targets, improving the reliability of the decision, but also enables the algorithm to adapt to different signal-to-noise ratio conditions and waveform parameters. Simultaneously, the mechanism of "retaining first, then discarding after all matches fail" avoids premature misjudgment of individual uncertain targets, improving the integrity of point cloud data and the system's fault tolerance.

[0083] The core of the defuzzification process in this invention lies in comparing velocity observations from two different types of frames: fast and slow. Due to the different velocity resolutions and fuzziness characteristics of the fast and slow frames, their "folding" patterns of the same true velocity in the velocity domain exhibit a fixed mathematical relationship (e.g., a 2x relationship). This invention transforms this waveform difference constraint into a computable, strongly constrained matching search problem by generating and comparing the velocity expansion matrix, thereby reliably calculating the unique true velocity.

[0084] Based on the same concept, the present invention also provides an electronic device, including a memory, a processor, and a computer program or instructions stored in the memory, wherein the processor executes the computer program or instructions to implement the high capture rate target detection method based on traffic radar as described above.

[0085] Based on the same concept, the present invention also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implements the high acquisition rate target detection method based on traffic radar as described above.

[0086] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0087] This invention systematically solves three core challenges in traffic radar detection through a complete and coherent signal processing chain: insufficient far-field angular resolution, inaccurate information due to velocity ambiguity, and difficulties in point cloud matching and fusion. The final output includes a target point cloud containing accurate distance, true velocity, and ultra-high resolution corrected angles, enabling high-precision synchronous measurement of all parameters of road target position, velocity, and azimuth, greatly enhancing the radar's perception capabilities.

[0088] This invention employs regional differential processing to perform road-prior-based focused beamforming in the far field (second region), effectively improving the signal-to-noise ratio and angular resolution of far-field targets. Combined with subsequent differential virtual array super-resolution processing, it further overcomes the aperture limitations of physical arrays, enabling clear differentiation and stable detection of small far-field targets and densely packed targets at close angles, thereby significantly improving the system's detection probability (capture rate) in key far-field regions.

[0089] This invention creatively introduces a processing flow of "generating velocity assumptions → Doppler phase compensation → parallel angle estimation → optimal angle spectrum decision," deeply integrating and simultaneously solving the two traditionally separate or sequential problems of velocity defuzzification and high-resolution angle estimation. Utilizing the quality (peak amplitude) of the angle spectrum as the decision criterion, the correct velocity is determined in reverse, achieving mutual verification and joint optimization of velocity and angle information. This fundamentally solves the subsequent angle estimation errors caused by velocity fuzziness, significantly improving the overall reliability of parameter estimation in complex scenarios.

[0090] This invention, through the design of "cross-frame matching based on precise distance and angle" and "de-ambiguity based on velocity extension matrix," can efficiently and accurately correlate and fuse point cloud data from alternating radar frames with different waveform characteristics. This not only outputs more accurate instantaneous velocity but also generates a stable target point cloud sequence that is spatiotemporally consistent and parameter-accurate, providing input data of far superior quality to traditional methods for subsequent target tracking, classification, and behavior analysis.

[0091] The entire scheme embodies the ideas of intelligent resource allocation and algorithmic collaboration: it avoids complex global calculations through region partitioning; it compensates for the limitations of hardware aperture through differential virtual array technology; and it improves parameter accuracy by utilizing time dimension information through sliding frame matching and deblurring. These designs enable the method of this invention to achieve a qualitative leap in detection performance without significantly increasing radar hardware costs and data processing power consumption, giving it extremely high engineering application value and market competitiveness. Attached Figure Description

[0092] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only one embodiment of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0093] Figure 1 This is a flowchart of a high acquisition rate target detection method based on traffic radar in an embodiment of the present invention;

[0094] Figure 2 This is a schematic diagram of the receiving antenna array and differential virtual array in an embodiment of the present invention.

[0095] Figure 3 This is a schematic diagram of angle measurement of the receiving antenna array in an embodiment of the present invention;

[0096] Figure 4 This is a schematic diagram of differential virtual array angle measurement in an embodiment of the present invention;

[0097] Figure 5 This is a schematic diagram of the radar beam arrangement sequence in an embodiment of the present invention. Detailed Implementation

[0098] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0099] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0100] Example 1

[0101] like Figure 1 As shown, the high acquisition rate target detection method based on traffic radar (such as traffic radar operating in the 90GHz band) provided in this embodiment of the invention includes the following steps:

[0102] Step S1: Based on the road geometry parameters, the detection matrix generated based on the radar echo signal is divided into a first region and a second region.

[0103] In the initial stage of the detection process, this invention first performs a crucial detection matrix region partitioning operation. The core objective of step S1 is to intelligently partition the entire detection matrix based on the prior road geometry information of the radar deployment scenario, so as to achieve optimized allocation of subsequent processing resources and accurate adaptation of algorithm strategies. The specific implementation process is as follows:

[0104] Step S1.1: Radar raw data preprocessing and detection matrix generation.

[0105] The raw echo signal received by the radar (i.e., the radar echo signal) is preprocessed. Specifically, for the multi-channel complex signal within each coherent processing interval, a range-dimensional Fast Fourier Transform (FFT) and a velocity-dimensional (Doppler-dimensional) FFT are performed sequentially. Through these two FFT processes, the time-domain echo signal is transformed into the range-Doppler two-dimensional frequency domain, generating a complete range-Doppler detection matrix. The size of the range-Doppler detection matrix is ​​mmdoppler × numsample, where mmdoppler is the number of points in the velocity dimension (Doppler dimension) and numsample is the number of points in the range dimension. Each cell (r, v) in the range-Doppler detection matrix contains signal energy (or complex amplitude) information at the range gate r and the velocity cell v.

[0106] Step S1.2: Set distance thresholds based on lane length.

[0107] The key to step S1.2 lies in utilizing prior knowledge of the road scenario, namely lane length. Lane length The lane length is set during system initialization based on the actual traffic scenario (such as road segment mode). The nearest end usually corresponds to the radar detection blind zone, and the farthest end is set to the radar's maximum effective detection range.

[0108] Based on lane length Calculate or directly set a distance threshold. In a preferred embodiment, a distance-based threshold is used. The subscript for the distance dimension is set to half the lane length. For example, if the lane length... If the distance is 50 meters and the distance resolution is 0.5 meters, then the distance division threshold is... It may be set to 50 (corresponding to 25 meters). Distance threshold. The physical significance of this is that it divides the radar's detection range into near-field and far-field regions in terms of distance. In the near-field region, target echoes are usually stronger but more complex in distribution, while in the far-field region, target echoes are weaker and more in line with road geometry constraints.

[0109] Step S1.3: Perform region division of the range-Doppler detection matrix.

[0110] Use the distance threshold determined in step S1.2 to divide the data. The range-Doppler detection matrix generated in step S1.1 is then divided as follows:

[0111] The first region is defined as all data units in the detection matrix whose range dimension subscript is less than or equal to the range division threshold L. This region corresponds to the near-field region. Due to the close target distance, the signal energy is usually strong, but it may be significantly affected by multipath propagation, obstruction, and other factors, resulting in a noticeable angular spread. Therefore, in this invention, instead of employing computationally complex beamforming for this region, efficient conventional CFAR detection and other processing methods are directly applied to quickly output reliable targets.

[0112] The second region is defined as all data units in the detection matrix whose distance dimension subscript is greater than the distance threshold L. This region corresponds to the far-field region. Target signal energy attenuation is severe in this region, making it a critical area for weak target detection. Furthermore, its azimuth distribution is more strictly constrained by lane geometry. Therefore, this invention will implement beamforming and focusing detection based on road geometry parameters in this region, which involves significant computational cost but yields substantial gains, to improve the detection probability and angle measurement accuracy of far-field weak targets.

[0113] Through steps S1.1 to S1.3 above, this invention completes the scene-adaptive detection matrix region partitioning, laying a crucial data foundation for subsequent differentiated, high-efficiency, and high-precision parallel processing. This partitioning method ensures that computing resources are prioritized for far-field weak target detection, where performance improvement is most needed, while avoiding unnecessary complex calculations in the near field, thus achieving an optimal balance between detection performance and system real-time performance overall.

[0114] Step S2: Perform beamforming and detection on the data in the second region based on road geometry parameters, and simultaneously detect the signal in the first region.

[0115] Step S2 performs parallel and differentiated detection processes on the two regions of the probe matrix to simultaneously ensure the efficiency of near-field processing and the performance of far-field detection. The specific implementation of step S2 includes the following two parallel processing branches:

[0116] Branch 1: Detect the signal in the first region (near field region).

[0117] From the first region obtained in step S1.3, the multi-channel complex signal corresponding to each range-Doppler cell is extracted directly. Instead of complex beamforming, the extracted multi-channel complex signals are directly subjected to constant false alarm rate (CFAR) detection on the signal of each receiving channel (or after incoherent energy accumulation of the multi-channel complex signals). This detection method has low computational complexity and high speed, and is suitable for near-field regions with high signal-to-noise ratios and strong target echoes. It aims to quickly and reliably output the range and velocity information (i.e., range-Doppler cell subscripts) of near-field targets.

[0118] Output: All point targets detected by CFAR (including distance dimension subscripts, velocity dimension subscripts, and preliminary amplitude information) are used as the detection results for the first region and temporarily stored for fusion.

[0119] Branch 2: Perform beamforming and detection on the data of the second region (far field region) based on road geometry parameters.

[0120] This branch is the core of improving the performance of far-field weak target detection. It is implemented for the second region, and the specific steps are as follows:

[0121] Step S2.1: Calculate the angle scanning range based on road geometry parameters.

[0122] Based on the known lane width The actual physical distance R corresponding to the preset distance threshold L. L (R) L =L×Δr, where Δr is the range resolution), to calculate the maximum scanning angle of beamforming. In a preferred embodiment, the maximum scanning angle is calculated using the following formula:

[0123] (1)

[0124] in, is the maximum scanning angle; atan is the arctangent function. The physical meaning of formula (1) is: at the actual physical distance R L At this point, the arctangent of the angle opened by half the lane width. Therefore, the angle scan range is limited to [ This ensures that the beam energy is focused within the effective area covered by the lane width, greatly suppressing clutter and interference outside the lane.

[0125] Step S2.2: Generate a set of discrete beam steering vectors.

[0126] Within the defined scanning range [ Within a given range, K discrete angle points are uniformly selected at preset angular intervals (e.g., 0.5° or 1°). }

[0127] For a radar receiving antenna array, the physical position of its nth element relative to the first element is known to be d. n (Using half-wavelength λ / 2 as the normalized unit, n = 1, 2, ..., , (Number of receiving channels). For each discrete angle point. Calculate its corresponding beam steering vector:

[0128] (2)

[0129] in, For the k-th discrete angle point The corresponding beam steering vector; j is the imaginary unit, and the superscript T indicates transpose. This yields a set of discrete beam steering vectors. .

[0130] Step S2.3: Perform beamforming on the signal in the second region.

[0131] For each distance-Doppler unit in the second region Extract its corresponding Channel complex signals (one (a column vector of ×1); using the beam steering vector generated in step S2.2, for the range-Doppler unit corresponding to... Channel complex signals Beamforming is performed. For the k-th beam direction, the synthesized output is:

[0132] (3)

[0133] in, This represents the beamforming result in the k-th beam direction, i.e., the complex signal in the k-th beam direction; the superscript H indicates the conjugate transpose. After calculating for all K beam directions, the complex signals of this range-Doppler element in the K beam directions are obtained. .

[0134] Repeat the above operation for all range-Doppler cells in the second region, thereby generating the detection matrix (size of mmdoppler×(numsample-L)×) corresponding to the second region. The transformation is into K beam domain detection matrices (each with a size of mmdoppler×(numsample-L)).

[0135] Step S2.4: Beam domain signal processing and detection.

[0136] For each element in each beam domain detection matrix Calculate its signal power Then, within a coherent processing interval, the signal power is... Perform incoherent accumulation (such as direct summation) to obtain each Energy value after unit accumulation Then, consider the energy distribution corresponding to the k-th beam direction. Independent CFAR detection is performed to detect potential target points in the beam direction. Each target point contains information: range subscript r, velocity subscript v, and beam direction angle. Accumulate energy value .

[0137] Since the same physical target may appear in multiple adjacent beam directions, the detection results of K beams need to be fused. For a specific... If a cell is detected in multiple beam directions, the energy value of that cell is selected for accumulation. The direction of the highest beam is used as the estimated angle of the target. Different The target of the unit is directly retained. All target points after fusion processing (including distance index, velocity index, preliminary estimated angle, and accumulated energy) are used as the detection results for the second region.

[0138] Step S3: Fuse the detection results of the second region and the first region to output a preliminary detection point cloud containing target distance and velocity information.

[0139] The detection results of the first region output from branch one and the detection results of the second region output from branch two in step S2 are merged to form a preliminary detection point cloud. For the very few possible instances where the same target is detected repeatedly at the boundary of two regions (with the same or very close distance and velocity indices), the point with the higher amplitude (or energy) after incoherent accumulation is selected as the output, while the other point is suppressed. Finally, the preliminary detection point cloud obtained after fusion contains the distance information (indices) and velocity information (indices) of all detected targets, as well as the preliminary estimated angle from the target in the second region, providing basic data for subsequent high-resolution angle correction and velocity deblurring processing.

[0140] Step S4: Based on the velocity information of each target in the preliminary detection point cloud and the radar's maximum unambiguous velocity, generate at least one velocity hypothesis, and based on the velocity hypothesis, generate multiple sets of candidate signals through Doppler phase compensation.

[0141] Step S4 is the key innovative step in this invention for achieving joint velocity and angle calculation. Its purpose is to generate a candidate true velocity hypothesis for each target that may experience velocity ambiguity, and to use this hypothesis to perform phase compensation on the original array signal, thereby providing multiple candidate, phase-consistent signal versions for subsequent high-resolution angle estimation. The specific implementation process is as follows:

[0142] Step S4.1: For each target in the preliminary detection point cloud, generate at least one velocity hypothesis.

[0143] For each target point i in the initial detection point cloud, obtain its key parameters:

[0144] Raw speed information This value originates from the initial detection point cloud and represents the velocity value corresponding to the target's index in its velocity dimension (Doppler dimension). This velocity information... It is possible that the speed is blurred and the image is folded in […]. ] Observation speed within the range.

[0145] Maximum unambiguous speed of radar This is a system constant determined by the current radar waveform parameters.

[0146] According to speed information The symbol generates a velocity hypothesis. This is considered the most likely candidate value for the true velocity of the target. The generation rules are as follows:

[0147] If speed information If >0, then the speed is assumed to be ;

[0148] If speed information If ≤0, then the velocity is assumed to be .

[0149] This generation rule is based on velocity ambiguity under TDMA (Time Division Multiple Access) waveforms with a resolution of 2. This is the physical principle of periodic folding. For an observed positive velocity... Its actual speed may be It may itself be (A smaller positive or negative velocity).

[0150] Step S4.2: For each velocity assumption, calculate its corresponding Doppler phase compensation value.

[0151] For the velocity assumption generated for each target Calculate the resulting Doppler phase difference (i.e., the Doppler phase compensation value) that needs to be compensated between the receiving channels:

[0152] (4)

[0153] in, Indicates velocity assumption The corresponding Doppler phase compensation value (unit: m / s); Indicates the center frequency of the current radar transmitted waveform (unit: Hz); Indicates the duration of a single transmitted pulse or the equivalent coherent processing time (unit: s); Represents the speed of light (unit: m / s). Formula (4) calculates the radial velocity of the target under the TDMA system. Additional phase terms introduced on channels with different combinations of transmit and receive antennas.

[0154] Step S4.3: Use the Doppler phase compensation value to compensate the original multi-channel received signal of the target and generate a candidate signal.

[0155] Step S4.3 is the core operation at the signal level, which generates multiple phase-corrected signal copies for each target.

[0156] Extracting the raw multi-channel received signal: Based on the distance index of target i in the preliminary detection point cloud. and speed subscript (Corresponding observation speed) ), extracting the [data] from the raw, unsynthesized range-Doppler detection matrix. The original multi-channel complex signal corresponding to the unit . It is a dimension × The column vector (for MIMO virtual arrays, and (representing the number of transmitting antennas and the number of receiving antennas, respectively), or simplified according to the processing flow. The receiving channel signal of the dimension.

[0157] The phase compensation value calculated in step S4.2 The extracted raw signal is applied. The above generates a phase-compensated candidate signal, using the following formula:

[0158] (5)

[0159] in, This represents the candidate signal after phase compensation; Indicates the transmit antenna index; This indicates the receiving antenna index.

[0160] The physical meaning of formula (5) is: if the true velocity of the target is exactly... Then this compensation will exactly offset the effect caused by that speed. Doppler phase difference introduced between receiving channels This makes the compensated signal In an ideal scenario, it is equivalent to an array signal of a stationary target or a target that has not experienced velocity ambiguity.

[0161] For the same target i, its original velocity It also implies a = The assumption is that the velocity is unambiguous. Therefore, for each target, we generate at least two sets of candidate signals:

[0162] Candidate signal group A: Based on velocity assumption (The signal obtained after phase compensation, generated from step S4.1) .

[0163] Candidate signal group B (optional but implicit): based on the original velocity (That is, the velocity assumption is) , correspond The signal obtained after phase compensation (calculation) .

[0164] Ultimately, each target corresponds to multiple sets (at least two sets) of candidate signals after phase compensation based on different velocity assumptions. , These candidate signals are fed into the subsequent step S5 for high-resolution angle estimation based on a differential virtual array.

[0165] Step S4 generates a velocity hypothesis using a concise rule and uses this hypothesis to perform precise phase compensation on the original array signal. This creatively transforms the problem of determining "which velocity is real" into a signal quality problem of "which set of phase-compensated signals can produce a better angular spectrum," providing a clear and calculable basis for subsequent joint decision-making.

[0166] Step S5: Perform high-resolution angle estimation processing on each group of candidate signals based on the principle of differential virtual array to obtain multiple angle spectra.

[0167] Step S5 receives multiple sets of candidate signals output from step S4 (each set corresponding to a velocity hypothesis), and uses Differential Virtual Array (DVA) technology to generate a high-resolution spatial angle spectrum for each set of signals, thereby providing a basis for subsequent selection of the correct velocity and angle. The specific implementation is as follows: the following process is executed independently for each set of candidate signals:

[0168] Suppose a non-uniform receiving antenna array has There are n array elements, and the physical position of the nth array element is a. n ×d, where n=1,2,..., d = λ / 2 is the reference antenna spacing, λ is the radar operating wavelength, and a n Let a be the relative position coefficient of the nth array element in units of d. For a uniform array, a n = n-1; for sparse arrays, a n This constitutes a non-uniform sequence.

[0169] Assume that the receiving antenna array simultaneously receives signals from M uncorrelated far-field targets. The incident signal vector is... The noise vector is The noise vector statistically follows a Gaussian distribution with a mean of 0 and a variance of . At this time, the receiving antenna array receives... ×1D quick capture signal It can be represented as:

[0170] (6)

[0171] in, for A ×M-dimensional guidance matrix; for the m-th direction... The signal, its corresponding ×1-dimensional guiding vector for:

[0172] (7)

[0173] For each group of candidate signals from step S4 (which can be viewed as a snapshot estimate on a specific distance-Doppler unit), The following steps S5.1 to S5.4 are executed independently:

[0174] Step S5.1: Calculate the spatial autocorrelation matrix of the candidate signal.

[0175] For a set of candidate signals for target i (one Estimate the spatial autocorrelation matrix of a 1×1 dimensional column vector:

[0176] (8)

[0177] in, This represents the estimated spatial autocorrelation matrix; This indicates the number of snapshots used for estimation, which in radar is typically the number of pulses within a coherent processing interval or the number of multichannel samples within a range-Doppler cell; the superscript H indicates a conjugate device. Let represent the candidate vector for the l-th snapshot. For point target detection, Typically, 1 or a small number is used, or a single snapshot estimate is used directly. .

[0178] The calculated spatial autocorrelation matrix It is × The Hermitian matrix. According to the signal model, the theoretical spatial autocorrelation matrix can be expressed as: (9)

[0179] in, This represents the theoretical spatial autocorrelation matrix; This indicates the operation of calculating the mean; for A ×M-dimensional guiding matrix; Indicates the m-th direction The guiding vector; This represents the autocorrelation matrix corresponding to the incident signal; Indicates the order is The identity matrix; This represents the noise power (i.e., variance) on each receiving channel. The covariance matrix represents the entire noise vector; Represents the m-th incident signal The power (variance).

[0180] Step S5.2: Vectorize the spatial autocorrelation matrix to obtain the initial differential signal vector.

[0181] The spatial autocorrelation matrix calculated in step S5.1 Perform vectorization operations:

[0182] (10)

[0183] The vec(·) operator converts the matrix... Stack them in columns to form a A column vector z of size ×1. According to the signal model, this vector z can be represented as:

[0184] (11)

[0185] in, for A matrix of size M, whose m-th column is... , Indicates taking the conjugate. This indicates the Kronecker product. That is, corresponding to the direction The equivalent steering vector of the differential virtual array; The signal power vector is M×1; It is A vector of size ×1, whose elements correspond to the matrix Except for the diagonal position which is 1, the rest are 0.

[0186] The vector z is the initial differential signal vector, which contains information about all the virtual array elements that can be generated from the original sparse array through differential operations.

[0187] Step S5.3: Remove redundancy and reorder the initial differential signal vector to obtain the equivalent uniform virtual array signal.

[0188] Step S5.3 extracts the usable, continuous virtual array signal from the theoretical vector z. Its core is to construct and apply the equivalent steering vector of the differential virtual array. .

[0189] according to , This can be further represented as a block structure as follows:

[0190] (12)

[0191] Among them, the nth sub-block (dimension is) The expression for ×1) is given by formula (13), which characterizes the phase relationship of the signal between array elements when the nth array element is taken as a reference, where the kth element depends on the relative position of array element k and the reference array element n. Based on d = λ / 2, The k-th element is:

[0192] (13)

[0193] in, This represents the position coordinates of the k-th array element; This represents the position coordinates of the nth specific reference array element.

[0194] The vector z obtained in step S5.2 has elements that are The structure corresponds to this. Each element in vector z is essentially mapped to a pair of physical array elements. The position of a difference virtual array element is determined by the index, and the coordinates of that position are... (in units of half wavelength).

[0195] Due to difference operations For different physical array element pairs It will be mapped to a position symmetrical about the origin. and- (This is because) different element pairs may map to the same difference position, resulting in a large number of duplicate (redundant) elements in the set of virtual elements corresponding to vector z. These redundant elements need to be processed. Specifically, for all elements in the z vector mapped to the same virtual element position, one value is selected and retained, or the arithmetic mean of the signal values ​​corresponding to all duplicate positions is taken as the final signal value at that unique virtual element position. This process removes redundant elements.

[0196] Reordering and extracting continuous uniform portions: The virtual array elements with unique positions obtained after redundancy removal are sorted according to their spatial coordinates ( The values ​​of the virtual array elements are sorted in ascending order. After sorting, the distribution of the virtual array element positions is observed. Typically, there is a continuous and uniformly distributed segment of virtual array elements near the zero point of the coordinate system, with an interval of half a wavelength. Generally, the longest continuous uniform virtual array element subset located at the center is selected for subsequent processing. Let this continuous uniform virtual array contain N... v Each array element.

[0197] Based on the continuous uniform virtual array element position sequence, the signal values ​​corresponding to these positions are extracted from the set of signal values ​​after redundancy has been removed, and arranged strictly according to their spatial position order to form an N-array. v ×1 column vector z virtual This z virtual This is the equivalent signal vector corresponding to a continuous uniform virtual array (i.e., the equivalent uniform virtual array signal). virtual The effective physical aperture is (N) v -1)×(λ / 2), significantly larger than the physical aperture of the original sparse array, such as Figure 2 As shown. Figure 2 In the middle, the top row is the receiving antenna array, and the bottom row is the differential virtual array. The original sparse array is reconstructed, typically by selecting the middle continuous array for subsequent angular spectrum estimation. The advantages of uniform array elements are used to suppress the sidelobes generated by the original sparse array, such as... Figure 3and Figure 4 As shown, the side lobes basically disappear, and only the peak value of the main lobe is used for angle measurement.

[0198] Step S5.4: Perform spatial spectrum estimation on the equivalent uniform virtual array signal to obtain the angular spectrum.

[0199] The equivalent uniform virtual array signal z obtained in step S5.3 is processed using a conventional beamforming method based on FFT. virtual Spatial spectrum estimation is performed. A conventional beamforming method based on FFT is used:

[0200] For z virtual Perform N FFT Point (N) FFT ≥N v Fast Fourier Transform (FFT); take the square of the FFT result to obtain the power spectrum; map the FFT points to spatial angles to obtain the discrete angle spectrum, which is a high-resolution angle spectrum corresponding to the candidate signal.

[0201] Step S5.5: Generate multiple angle spectra for each target.

[0202] For each set of candidate signals for the same target i, repeat steps S5.1 to S5.4. Ultimately, each target will obtain multiple angular spectra.

[0203] Step S5 uses differential virtual array technology to "transform" the observations from the sparse array into signals from an equivalent large-aperture uniform array, and then performs high-resolution spectral estimation on them. This process generates an independent angular spectrum for each velocity hypothesis, thus transforming the joint estimation problem of "velocity-angle" into a problem of comparing the quality of different angular spectra.

[0204] Step S6: Calculate the peak amplitude of each angle spectrum, and take the angle value corresponding to the angle spectrum with the largest peak amplitude as the target correction angle, thereby obtaining a detection point cloud containing target distance, velocity information and correction angle.

[0205] The purpose of step S6 is to analyze the quality of multiple angle spectra generated from different velocity assumptions for the same target (using peak amplitude as the primary criterion) and select the set that best matches the real physical scenario, thereby simultaneously determining the final accurate correction angle and the true velocity for de-ambiguation of the target. The specific implementation process is as follows:

[0206] Step S6.1: Extract the peak amplitude and corresponding angle of multiple angular spectra for each target.

[0207] For the same target i from step S5 (whose preliminary information includes the distance index r) i Observation velocity subscript v iPerform the following operations on the multiple angle spectra of ) :

[0208] For each angular spectrum, find the amplitude and angle values ​​corresponding to its global maximum value (peak value) within its effective angular scanning range. Create a list for target i, recording the peak information for each angular spectrum: (spectral index, peak amplitude, peak-corresponding angle, velocity assumption corresponding to the spectrum).

[0209] Step S6.2: Make a joint decision based on the peak amplitude to determine the final angle and velocity.

[0210] A correct velocity assumption will, through precise phase compensation, enable the energy of the array signal to achieve the most ideal in-phase superposition in the direction of the real target, thereby producing a sharpest and highest amplitude main lobe peak in its angular spectrum.

[0211] Compare all the peak amplitudes recorded for target i, select the angle spectrum with the largest peak amplitude, and obtain the spectral index, the angle corresponding to the peak, and the velocity hypothesis corresponding to the spectrum. Among them, the angle corresponding to the peak is the high-precision angle estimate (i.e., the corrected angle) of the target obtained after super-resolution processing and Gaussian spectral quality judgment.

[0212] If the velocity corresponding to this spectrum is assumed to be equal to the original observed velocity. If the target speed is not blurred, then the true speed is determined to be... If the velocity corresponding to this spectrum is assumed to be... Through ±2 If the velocity assumption generated by the rule is ambiguous, then the target velocity is determined to be the velocity assumption corresponding to that spectrum. The target's distance information r i It remains unchanged.

[0213] For each target in the initial detection point cloud, repeat steps S6.1 to S6.2. Refine the point cloud entries (distance information r) for all targets. i (The velocity information and the corrected angle) are collected to form a new point cloud set.

[0214] Step S6 intelligently selects the most reasonable combination of velocity and angle by comparing the indirect method of "signal restoration quality" (angle spectrum peak value) under different velocity assumptions. This method integrates and solves the two problems of velocity deblurring and angle super-resolution simultaneously, making full use of all the information of the array signal, and finally outputs a target point cloud with highly accurate parameters and strong consistency, laying the optimal data foundation for subsequent cross-frame matching and tracking.

[0215] Step S7: Based on the target distance and correction angle, perform cross-frame matching and velocity deblurring on the fast and slow frame point clouds generated by the radar in a specific time sequence with different beam types, and output the final target point cloud with accurate parameters after fusion.

[0216] The purpose of step S7 is to utilize the point cloud data from subframes with different waveform characteristics emitted alternately by the radar, and through spatiotemporal correlation and velocity deblurring algorithms, to generate a final target list containing high-precision distance, angle, and the true velocity after deblurring. The specific implementation process is based on a specific beam scheduling strategy and matching deblurring algorithm, as follows:

[0217] Step S7.0: Beam scheduling and data caching strategy.

[0218] The radar beams are transmitted alternately in a fixed sequence of "narrow beam fast frame, wide beam slow frame, narrow beam slow frame, wide beam fast frame" to form a complete cycle, such as... Figure 5 As shown. Figure 5 In the text, Frame1~Frame6 represent frames 1 to 6, and chirp represents the linear frequency modulation signal.

[0219] Buffer initialization: Maintain a data buffer to cache historical subframe point cloud data that has been processed.

[0220] Sliding frame processing logic: Receive and process the kth subframe (k is the frame number).

[0221] The matching object is dynamically determined based on the beam type (wide / narrow) and frame type (fast / slow) of the current subframe:

[0222] If the current frame is a wide-beam slow frame, then select cached wide-beam fast frame point cloud data with the same beam type (wide) but different frame type (fast) from the buffer for matching; if the current frame is a narrow-beam fast frame, then select cached narrow-beam slow frame point cloud data with the same beam type (narrow) but different frame type (slow) from the buffer for matching.

[0223] This "out-of-order arrangement and sliding frame matching" strategy enables the system to complete a "fast-slow frame" matching of a wide beam and a narrow beam within any time window (such as 80ms), achieving efficient point cloud output at a fixed refresh rate and optimizing the 160ms delay required by traditional sequential processing to 80ms.

[0224] Step S7.1: Obtain and prepare matching point cloud data.

[0225] Current subframe point cloud: Obtain the refined detection point cloud corresponding to the current subframe (the kth frame) output in step S6, where the data for each target point i are the distance index, velocity index, and correction angle.

[0226] Historical subframe point cloud (frame k-2): According to the rules in step S7.0, the corresponding historical frame refined detection point cloud is retrieved from the buffer.

[0227] Step S7.2: Coarse matching of point clouds based on weighted nearest neighbor algorithm.

[0228] Step S7.2 aims to quickly identify point pairs from two point cloud frames that are most likely to correspond to the same physical target in terms of spatial location (distance, angle). Specifically, this includes:

[0229] Step S7.21: Calculate the normalized distance difference and normalized angle difference between the i-th point in the current subframe point cloud and the j-th point in the historical subframe point cloud.

[0230] (14)

[0231] (15)

[0232] in, These represent the distance indices of the i-th point in the current subframe point cloud and the j-th point in the historical subframe point cloud, respectively. This represents the maximum value in the distance dimension; These represent the angle values ​​of the i-th point in the current subframe point cloud and the j-th point in the historical subframe point cloud, respectively. Indicates the maximum field of view; This represents the normalized distance difference; This represents the difference in normalized angles.

[0233] Step S7.22: Calculate the weighted fusion value based on the normalized distance difference and the normalized angle difference. The specific formula is as follows:

[0234] (16)

[0235] in, Indicates the weighted fusion value; and This represents the weighting coefficients for the normalized distance difference and the normalized angle difference.

[0236] Step S7.23: Compare the weighted fusion value with the preset matching threshold.

[0237] If the weighted fusion value If the value is less than the preset matching threshold, the i-th point in the current subframe point cloud and the j-th point in the historical subframe point cloud are considered to be spatially close enough to be two observations of the same target, and the match is successful, proceeding to velocity deblurring processing. Otherwise, the match fails.

[0238] Step S7.3: Velocity defuzzification processing based on the velocity extension matrix.

[0239] For each coarsely matched point cloud pair (i,j), precise velocity calculation is performed using the different velocity resolution characteristics of fast and slow frames. Let the maximum value of the velocity dimension index in both fast and slow frames be... (Determined by waveform parameters), a preset even-number expansion factor P (e.g., P=4).

[0240] Generate the velocity index v of the i-th point in the target point cloud data of the current subframe. k Extended matrix :

[0241] (17)

[0242] Generate the velocity index v of the j-th point in the historical subframe point cloud. j Extended matrix :

[0243] (18)

[0244] These two extended matrices enumerate points i and j respectively, considering velocity ambiguity (in 2*). In the case of a period, all possible corresponding real velocity indices.

[0245] Calculate the extended matrix and Find the absolute value of the difference between all pairs of elements in the algorithm, and then find the minimum value among them. If the minimum value is less than a preset speed threshold, the speed match is considered successful. At this point, the value corresponding to the minimum value... The element in the matrix represents the correct velocity index of point i after defuzzification. Based on the minimum value... The elements in the calculation yield the target's true radial velocity.

[0246] If the minimum value is greater than or equal to the preset speed threshold, the speed matching is determined to be unsuccessful. Point i in the current subframe is temporarily retained and attempts to match it with other points in the historical subframe are continued. If point i in the current subframe fails to match with any points j in the historical subframe, point i is discarded after all matches in the current subframe have been processed.

[0247] S7.4: Fusion output final target point cloud

[0248] For each point pair (i,j) that successfully passes coarse matching in step S7.2 and velocity matching in step S7.3, generate a final target entry:

[0249] Distance: take r i or (r) i +r j) / 2 (or choose the better one based on the signal-to-noise ratio), and convert it to physical distance.

[0250] Angle: Use the corrected angle from the refined point cloud.

[0251] Speed: The actual speed calculated using S7.3.

[0252] All successfully matched and solved final target entries within this period (e.g., within an 80ms window) are aggregated to form the final target point cloud.

[0253] The refined point cloud of the current subframe is stored in the buffer according to its beam type, replacing the old frame data of the same type, in order to prepare for matching in the next subframe.

[0254] Step S7 efficiently integrates the coverage advantages of wide and narrow beams and the velocity measurement advantages of fast and slow frames through innovative sliding frame scheduling and "same beam, different frame" matching rules. Weighted matching using precise correction angles significantly improves correlation accuracy; furthermore, velocity ambiguity between fast and slow frames is robustly resolved through velocity expansion matrix comparison. The final output is a highly accurate and spatiotemporally consistent fused point cloud in both spatial location and velocity dimensions, greatly improving the radar system's target acquisition rate and parameter measurement accuracy, providing high-quality input for upper-level perception and decision-making.

[0255] Example 2

[0256] This invention also provides an electronic device, which includes a memory, a processor, and a computer program or instructions stored in the memory. The processor executes the computer program or instructions to implement the high capture rate target detection method based on traffic radar according to this invention.

[0257] Although not shown, the electronic device includes a processor that can perform various appropriate operations and processes based on programs and / or data stored in read-only memory (ROM) or loaded from a storage portion into random access memory (RAM). The processor can be a multi-core processor or may contain multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more specialized coprocessors, such as a central processing unit, graphics processing unit (GPU), neural network processor (NPU), digital signal processor (DSP), etc. Various programs and data required for device operation are also stored in RAM. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0258] The processor and memory described above are used together to execute programs / instructions stored in the memory. When the program / instructions are executed by the computer, they can implement the methods, steps, or functions described in the above embodiments.

[0259] Although not shown, embodiments of the present invention also provide a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implements the high acquisition rate target detection method based on traffic radar in embodiments of the present invention.

[0260] Readable storage media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0261] The above description only discloses specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or modifications that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A traffic radar-based high capture rate target detection method, characterized by, The detection method comprises: According to the road geometric parameters, the detection matrix generated based on the radar echo signal is divided into a first region and a second region; The data in the second region is subjected to beam synthesis and detection based on the road geometric parameters, and the signals in the first region are subjected to detection; The detection results of the second region and the first region are fused, and a preliminary detection point cloud containing target distance and speed information is output; Based on the speed information of each target in the preliminary detection point cloud and the maximum unambiguous speed of the radar, at least one speed hypothesis is generated, and based on the speed hypothesis, a plurality of groups of candidate signals are generated through Doppler phase compensation; Each group of candidate signals is subjected to high-resolution angle estimation processing based on the principle of differential virtual array to obtain a plurality of angle spectra; The peak amplitudes of each angle spectrum are calculated, and the angle value corresponding to the angle spectrum with the maximum peak amplitude is taken as the corrected angle of the target, and a detection point cloud containing target distance, speed information and corrected angle is obtained; Based on the target distance and the corrected angle, the fast and slow frame point clouds of different beam types alternately generated by the radar are subjected to cross-frame matching and speed unblurring processing, and a final target point cloud with accurate parameters is output after fusion. 2.The traffic-radar-based high-capture-rate target detection method of claim 1, wherein, According to the road geometric parameters, the detection matrix generated based on the radar echo signal is divided into a first region and a second region, comprising: The radar echo signal is subjected to Fourier transform in the distance dimension and the speed dimension to generate a range-Doppler detection matrix; According to the lane length in the road geometric parameters, a distance division threshold is set; The region with a distance dimension subscript less than or equal to the distance division threshold in the range-Doppler detection matrix is divided into the first region, and the region with a distance dimension subscript greater than the distance division threshold is divided into the second region. 3.The traffic-radar-based high-capture-rate target detection method of claim 1, wherein, The data in the second region is subjected to beam synthesis and detection based on the road geometric parameters, comprising: Based on the lane width in the road geometric parameters and a preset distance division threshold, the angle scanning range is calculated; According to the angle scanning range, a set of discrete beam steering vectors are generated; The multi-channel complex signals extracted from the second region are subjected to beam synthesis using the beam steering vectors to obtain complex signals in multiple beam directions; The complex signals in each beam direction are subjected to modulus square operation to obtain signal power; The signal power is subjected to non-coherent accumulation within a coherent processing interval to obtain an accumulated energy distribution; The accumulated energy distribution is subjected to constant false alarm rate detection to output the target detection result of the second region.

4. The high capture rate target detection method based on traffic radar according to claim 3, characterized in that, According to the angle scanning range, a set of discrete beam steering vectors are generated, comprising: K discrete angle points are selected at a preset interval within the angle scanning range; According to the physical positions of the radar receiving antenna array, the beam steering vector of each discrete angle point is calculated. 5.The traffic-radar-based high-capture-rate target detection method of claim 1, wherein, Based on the speed information of each target in the preliminary detection point cloud and the maximum unambiguous speed of the radar, at least one speed hypothesis is generated, comprising: acquiring speed information of each target in the preliminary detection point cloud and maximum unambiguous velocity of the radar ; generating speed information for each target in the point cloud based on the preliminary detection and the maximum unambiguous velocity of the radar generating the at least one speed hypothesis If the speed information is greater than 0, the speed hypothesis is ; If the speed information is less than or equal to 0, the speed hypothesis is ; and the speed information of each object in the preliminary detection point cloud directly as the speed hypothesis. 6.The traffic-radar-based high-capture-rate target detection method of claim 1, wherein, Based on the speed hypothesis, a plurality of groups of candidate signals are generated through Doppler phase compensation, comprising: For each speed hypothesis generated for each target, a corresponding Doppler phase compensation value is calculated; Using each of the Doppler phase compensation values, the original multi-channel received signal is phase compensated respectively; Wherein each of the phase-compensated signals constitutes a candidate signal, and the original multi-channel received signal is a signal extracted from the detection matrix according to the distance and speed information of the target. 7.The traffic-radar-based high-capture-rate target detection method of claim 1, wherein, For each candidate signal, high-resolution angle estimation processing based on the principle of differential virtual array is performed, including: For each candidate signal, a corresponding spatial autocorrelation matrix is calculated; A vectorization operation is performed on the spatial autocorrelation matrix to obtain an initial differential signal vector; An element position de-redundancy and reordering processing is performed on the initial differential signal vector to obtain a corresponding equivalent uniform virtual array signal; A spatial spectrum estimation is performed on the equivalent uniform virtual array signal to obtain an angle spectrum corresponding to the candidate signal.

8. The high capture rate target detection method based on traffic radar according to any one of claims 1-7, characterized in that, Based on the target distance and the corrected angle, the fast and slow frame point clouds of different beam types alternately generated by the radar in a specific time sequence are cross-frame matched and velocity de-masking processed, including: A buffer is maintained for buffering target point cloud data of at least one historical subframe, wherein the target point cloud data at least includes the distance index, the corrected angle and the speed index of the target; The target point cloud data of the current subframe is obtained, and from the buffer, the target point cloud data of a historical subframe with the same beam type and different frame type and the closest neighbor of the current subframe is selected for matching; Based on the distance index and the corrected angle, the target point cloud data of the current subframe and the historical subframe is matched; For the matched point cloud pair, velocity de-masking processing is performed based on the speed index to determine the real speed of the target. 9.The high capture rate target detection method based on traffic radar according to claim 8, wherein, The weighted nearest neighbor algorithm is used to match the target point cloud data of the current subframe and the historical subframe, including: The normalized distance difference and the normalized angle difference between the i th point in the target point cloud data of the current subframe and the j th point in the target point cloud data of the historical subframe are calculated respectively; A weighted fusion value is calculated according to the normalized distance difference and the normalized angle difference; If the weighted fusion value is less than a preset matching threshold, it is determined that the i th point and the j th point are matched successfully, and the velocity de-masking processing is entered. 10.The traffic-radar-based high-capture-rate target detection method of claim 8, wherein, For the matched point cloud pair, velocity de-masking processing is performed based on the speed index, including: For the matched point cloud pair, a speed expansion matrix of the i th point is generated according to the speed index, the maximum speed index and a preset expansion multiple of the i th point in the target point cloud data of the current subframe; a speed expansion matrix of the j th point is generated according to the speed index, the maximum speed index and a preset expansion multiple of the j th point in the target point cloud data of the historical subframe; The absolute value of the difference between all elements of the speed expansion matrix of the i th point and the speed expansion matrix of the j th point is calculated, and the minimum value is selected; If the minimum value is less than a preset speed threshold, it is determined that the point cloud pair is matched successfully, and the speed corresponding to the speed index in the speed expansion matrix of the i th point corresponding to the minimum value is the real speed. If the minimum value is greater than or equal to a preset speed threshold value, it is determined that the point cloud fails in speed matching, and the point is reserved first until the i th point in the target point cloud data of the current subframe and all points in the target point cloud data of the historical subframe fail in speed matching, and the i th point is discarded.

11. An electronic device comprising a memory, a processor, and a computer program or instructions stored on the memory, wherein, The processor executes the computer program or instruction to implement the traffic radar-based high-capture-rate target detection method in any one of claims 1-10.

12. A computer readable storage medium having stored thereon a computer program or instructions, characterized in that, The computer program or instruction is executed by the processor to implement the traffic radar-based high-capture-rate target detection method in any one of claims 1-10.

Images