CN-121165059-B - Multi-path false target detection and identification method for vehicle-mounted millimeter wave radar

CN121165059BCN 121165059 BCN121165059 BCN 121165059BCN-121165059-B

Abstract

The embodiment of the application provides a multipath false target detection and identification method of a vehicle millimeter wave radar. The method comprises the steps of obtaining a single-frame echo signal of a target vehicle millimeter wave radar, screening and filtering the single-frame echo signal to generate a single-frame point trace set, carrying out multipath rule judgment on the single-frame point trace set to generate a multipath judgment result, carrying out track correlation on single-frame passing point traces in the single-frame point trace set based on the multipath judgment result and a preset rule to generate a target track, carrying out multipath confidence information calculation processing on the target track to generate a comprehensive multipath confidence evaluation value, and carrying out multipath false target detection and identification processing based on the comprehensive multipath confidence evaluation value and a preset confidence threshold. In this way, multipath judgment can be accurately performed under actual complex road conditions.

Inventors

AN QINGRU
XU HAORAN
QIU LING
LI RONGLIN
LI BIAO
ZHANG WENCHAO

Assignees

北京理工睿行电子科技有限公司

Dates

Publication Date: 20260505
Application Date: 20251117

Claims (8)

1. The method for detecting and identifying the multipath false target of the vehicle millimeter wave radar is characterized by comprising the following steps of: acquiring a single-frame echo signal of a target vehicle millimeter wave radar; screening and filtering the single-frame echo signals to generate a single-frame point trace set; Judging whether the single-frame detected point trace in the single-frame point trace set meets a preset multipath rule equation or not; If the single-frame passing detection point meets a preset multipath rule equation, judging the single-frame passing detection point as a multipath false target and deleting the multipath false target; if the single-frame passing detection point does not meet a preset multipath rule equation but is within the redundancy range of the preset multipath rule equation, giving a multipath confidence to the single-frame passing detection point; Performing track association on the single-frame passing point track in the single-frame point track set based on the multipath judging result and a preset rule to generate a target track; Calculating the arithmetic average value of the multipath confidence coefficient of all the relevant point tracks of the target track in a single frame, and taking the arithmetic average value as the track confidence coefficient of the single frame passing point track; Carrying out exponential weighted average calculation on the track confidence coefficient of continuous multiframes to obtain a comprehensive multipath confidence coefficient evaluation value; and carrying out multipath false target detection and identification processing based on the comprehensive multipath confidence evaluation value and a preset confidence threshold value.
2. The method of claim 1, wherein the filtering the single-frame echo signal to generate a single-frame trace point set comprises: performing Fourier transform processing on the single-frame echo signals to generate a two-dimensional frequency domain matrix; detecting constant false alarm rate in the two-dimensional frequency domain matrix, and screening to obtain a single frame of detection point trace; and calculating target calculation data of each single-frame passing point trace, and sequencing the single-frame passing point traces according to the point trace distance to generate a single-frame point trace set, wherein the target calculation data comprises distance, speed, azimuth angle and energy value.
3. The method of claim 2, wherein said determining whether the single-frame overstock in the set of single-frame trails satisfies a preset multipath law equation comprises: acquiring all judgment conditions of the preset multipath rule equation; judging whether the single frame passing detection point meets all the judging conditions; If the single-frame passing detection point meets all the judging conditions, judging that the single-frame passing detection point meets a preset multipath rule equation; And if the single-frame passing point does not meet all the judging conditions, judging that the single-frame passing point does not meet a preset multipath rule equation.
4. The method of claim 3, wherein said assigning a multipath confidence to said single frame overstock comprises: when the judging condition coincidence quantity of the single frame passing detection point is a first data value; performing relaxation threshold processing on the single-frame passing detection point based on the target calculation data, and judging whether the single-frame passing detection point subjected to the relaxation threshold processing meets all the judging conditions; If the single frame passing detection trace processed by the widening threshold meets all the judging conditions, the multi-path confidence degree endowed to the single frame passing detection trace is a first confidence degree; if the single frame passing detection trace processed by the widening threshold does not meet all the judging conditions, the multipath confidence degree endowed to the single frame passing detection trace is a second confidence degree; When the judging condition of the single frame passing detection point accords with the second data value; performing relaxation threshold processing on the single-frame passing detection point based on the target calculation data, and judging whether the single-frame passing detection point subjected to the relaxation threshold processing meets all the judging conditions; If the single frame passing detection trace processed by the widening threshold meets all the judging conditions, the multi-path confidence degree endowed to the single frame passing detection trace is a third confidence degree; If the single frame passing detection trace processed by the widening threshold does not meet all the judging conditions, the multipath confidence degree endowed to the single frame passing detection trace is a fourth confidence degree; And when the judging condition coincidence quantity of the single frame passing detection point is a third data quantity value, the multipath confidence degree endowed to the single frame passing detection point is a fifth confidence degree.
5. The method of claim 1, wherein the performing multipath false object detection recognition processing based on the integrated multipath confidence score and a preset confidence threshold comprises: comparing the comprehensive multipath confidence evaluation value with a preset confidence threshold value, and judging whether the comprehensive multipath confidence evaluation value is not smaller than the preset confidence threshold value; if the comprehensive multipath confidence coefficient evaluation value is not smaller than the preset confidence coefficient threshold value, judging that the target track is a multipath false target track and eliminating the multipath false target track; and if the comprehensive multipath confidence evaluation value is smaller than the preset confidence threshold value, judging that the target track is a real track and reserving reporting.
6. The utility model provides a multipath false target detection recognition device of on-vehicle millimeter wave radar which characterized in that includes: the echo signal acquisition module is used for acquiring a single frame echo signal of the target vehicle millimeter wave radar; the point trace set generating module is used for screening and filtering the single-frame echo signals to generate a single-frame point trace set; The multi-path judging module is used for judging the multi-path rule of the single-frame point trace set and generating a multi-path judging result; If the single-frame passing detection point meets a preset multipath rule equation, judging the single-frame passing detection point as a multipath false target and deleting the multipath false target; if the single-frame passing detection point does not meet a preset multipath rule equation but is within the redundancy range of the preset multipath rule equation, giving a multipath confidence to the single-frame passing detection point; the target track generation module is used for carrying out track association on the single-frame passing detection point track in the single-frame point track set based on the multipath judgment result and a preset rule to generate a target track; The comprehensive evaluation generation module is used for carrying out multipath confidence information calculation processing on the target track to generate a comprehensive multipath confidence evaluation value; calculating the arithmetic average value of the multipath confidence coefficient of all the associated point tracks of the target track in a single frame, and taking the arithmetic average value as the track confidence coefficient of the single frame passing point track; Carrying out exponential weighted average calculation on the track confidence coefficient of continuous multiframes to obtain a comprehensive multipath confidence coefficient evaluation value; And the false target recognition module is used for carrying out multipath false target detection recognition processing based on the comprehensive multipath confidence evaluation value and a preset confidence threshold value.
7. An electronic device comprising a memory and a processor, the memory having a computer program stored thereon, characterized in that the processor, when executing the computer program, implements the method according to any of claims 1-5.
8. A computer readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the method according to any one of claims 1-5.

Description

Multi-path false target detection and identification method for vehicle-mounted millimeter wave radar Technical Field The embodiment of the application relates to the field of intelligent driving environment sensing, in particular to a multipath false target detection and identification method of a vehicle millimeter wave radar. Background The vehicle millimeter wave radar has the advantages of all-weather operation, accurate speed measurement, moderate cost and the like, and becomes the standard configuration of an ADAS/autopilot system. According to SAEJ3016 standard, the above-L2-level automatic driving system needs to be at least configured with 3-5 millimeter wave radars to realize 360-degree environment perception. However, when radar waves encounter strong reflectors such as bridge guardrails, tunnel walls, metal signboards and the like in the propagation process, multipath interference is generated, and common multipath mainly comprises false signals formed by return of signals after the signals indirectly reach a real target through the reflectors and multiple reflections of the signals between the reflectors and the target, namely multipath false signals. Multipath false belongs to millimeter wave radar target detection pain points, false alarm and false braking can be caused in actual road conditions, and therefore intelligent driving experience is reduced, and the problem is solved. In order to solve the existing problems, the prior art mostly adopts a training data set based on a deep neural network or acquires track information of a target to be judged received by a millimeter wave radar, but the existing scheme has a great number of defects, for example, the scheme of the training data set depends on a great number of labeling data, the acquisition cost of a multi-path scene truth value tag is high, the training set is extremely complex due to the fact that the diversity and the diversity of a complex scene of actual road conditions are large, and meanwhile, the memory occupation is too high, so that the scheme cannot be applied to the actual scene, or the scheme of the track information utilizes the track relation between the real target and a multi-path false target caused by a stationary point reflector to match the real target with the false target, so that the recognition of the multi-path false target is realized, if the initial clustering of points is wrong or the multi-path points are combined with other targets, the multi-path analysis of the multi-path is invalid, and therefore, a technology capable of accurately carrying out multi-path judgment under the complex road conditions is needed. Disclosure of Invention According to the embodiment of the application, the method for detecting and identifying the multipath false target of the vehicle-mounted millimeter wave radar is provided, and the multipath judgment can be accurately carried out under actual complex road conditions. In a first aspect of the application, a method for detecting and identifying multipath false targets of a vehicle-mounted millimeter wave radar is provided. The method comprises the following steps: acquiring a single-frame echo signal of a target vehicle millimeter wave radar; screening and filtering the single-frame echo signals to generate a single-frame point trace set; performing multipath rule judgment on the single-frame point trace set to generate a multipath judgment result; Performing track association on the single-frame passing point track in the single-frame point track set based on the multipath judging result and a preset rule to generate a target track; performing multipath confidence information calculation processing on the target track to generate a comprehensive multipath confidence evaluation value; and carrying out multipath false target detection and identification processing based on the comprehensive multipath confidence evaluation value and a preset confidence threshold value. In one possible implementation manner, the filtering the single-frame echo signal to generate a single-frame point track set includes: performing Fourier transform processing on the single-frame echo signals to generate a two-dimensional frequency domain matrix; detecting constant false alarm rate in the two-dimensional frequency domain matrix, and screening to obtain a single frame of detection point trace; and calculating target calculation data of each single-frame passing point trace, and sequencing the single-frame passing point traces according to the point trace distance to generate a single-frame point trace set, wherein the target calculation data comprises distance, speed, azimuth angle and energy value. In one possible implementation manner, the performing multipath rule judgment on the single frame point trace set, and generating a multipath judgment result includes: Judging whether the single-frame passing point trace in the single-frame point trace set meets a preset multipath rule equation or not;