A wheel feature recognition method

Abstract

The present application relates to a wheel feature recognition method for use in automotive electronic products, the method comprising: S1, determining a Doppler dimension search range for a target wheel and obtaining an initial signal of the target wheel within the search range; S2, performing constant false alarm processing on the Doppler dimension of the distance term of the initial signal to obtain a target array; S3, obtaining a target number of points by accumulating each element of the target array; S4, determining whether the target number of points is greater than or equal to a preset number of tire points; S5, if so, obtaining a wheel feature array; S6, otherwise repeating steps S1-S4 to fill and iterate the target array. The beneficial effect of the present application is that the present application effectively eliminates false target points generated by tires, preventing interference with trajectories during data processing.

Description

Technical Field

[0001] The present application relates to the field of automotive electronics technology, and more specifically, to a wheel feature recognition method. Background Art

[0002] As autonomous driving levels continue to rise, radar chip performance continues to improve, along with range and velocity resolution, resulting in a stronger radar's ability to distinguish objects. In the automotive radar field, close-range detection of moving objects is often hampered by factors such as low-frequency envelopes, vehicle noise, and target and environmental interference, leading to significant errors in speed and angle measurements. A particularly serious issue is interference from the target vehicle's tires. Because wheels rotate about their axles, the speed of each point on the wheel in the radar coordinate system varies, creating numerous interference points that disrupt radar signal processing. Once these interference points are fed into the data processing, they significantly impact the radar sensor's trajectory recognition. Due to the high level of signal processing integration, the commonly used radar waveform is the Fast Chirp variant of the FMCW waveform. Signals are processed using a radar cube formed from two dimensions: range and velocity. Many signal processing algorithms are pre-installed in the processor for hardware acceleration, and existing signal processing algorithms are generally not significantly modified. Therefore, specialized tire optimization cannot be performed at the signal processing level, and the resulting interference cannot be eliminated at the signal processing end.

[0003] In close-range scenarios detected by vehicle-mounted radar, tires cause strong interference to measurement results, which cannot be completely eliminated. This will affect subsequent data processing and increase the probability of false alarms and missed alarms of the radar. Summary of the Invention

[0004] In order to overcome the above-mentioned problems in the prior art, the present application provides a wheel feature recognition method.

[0005] A wheel feature recognition method, applied to automotive electronic products, comprises:

[0006] S1, determining the Doppler search range of the target wheel and obtaining the initial signal of the target wheel within the search range;

[0007] S2, performing constant virtual warning processing on the range-Doppler dimension of the initial signal to obtain a target array;

[0008] S3, obtaining the target points by accumulating the elements of the target array;

[0009] S4, determining whether the target number of points is greater than or equal to the preset number of tire points;

[0010] S5, if based on the target number of points, obtain the wheel feature array;

[0011] S6, otherwise repeat steps S1-S4 to iterate and fill the target array.

[0012] Optionally, in step S1, determining the Doppler search range of the target wheel includes:

[0013] The number of sampling points n spanned by the wheel in the distance dimension is determined by the distance resolution, and the data of n distance items are used as the search range of the wheel data for searching;

[0014] The search range is determined by the following formula: Where l is the wheel diameter.

[0015] Optionally, step S2 includes:

[0016] The initial signal is sequentially subjected to sliding window selection processing to select target points, and the target points are subjected to peak screening to obtain a target array.

[0017] Optionally, the step of sequentially performing a sliding window selection process on the initial points to select a target point includes:

[0018] A protection unit is set on two adjacent sides of the inspected object, and the selected area on both sides of the protection unit is divided into a left window X and a right window Y;

[0019] Perform sum and average calculation on each point of the left window X and the right window Y, compare the amplitude of the object to be inspected with a preset adaptive threshold, and obtain the target point.

[0020] Optionally, the preset adaptive threshold is any one of an average value of the left window X and the right window Y, a smaller value of the left window X and the right window Y, and a larger value of the left window X and the right window Y.

[0021] Optionally, the step of filtering the target points by peak values ​​to obtain a target array includes:

[0022] Delete the results of the surrounding points whose amplitude is higher than the target point, and store the remaining target points in the target array ; Target array Among them, a i The larger it is, the more targets there are at the same distance but different speeds, and the more similar it is to the wheel state.

[0023] Optionally, in step S5, obtaining a wheel feature array according to the target number of points includes:

[0024] Get the current distance item number γ and calculate the target wheel position k. The formula is: in, R max is the maximum recognition distance;

[0025] The distance item of the wheel target position k is set as the tire feature observation center point, the tire feature data K is stored and the target array is cleared.

[0026] Optionally, repeating steps S2 to S5, after all initial signals are subjected to constant false alarm processing, if the tire characteristic data K is not empty, obtaining tire characteristic points according to the tire characteristic data K.

[0027] Optionally, obtaining tire feature points according to the tire feature data K specifically includes the following steps:

[0028] S7, the tire characteristic observation center point k of the tire characteristic data K i As the wheel center, calculate the longitudinal distance: y j =k i *ΔR*cos(θ j )…(4), the vertical distance range is or The points h inside are filtered out;

[0029] S8, if point h is greater than b, calculate the horizontal distance and direction by angle: x j =k i *ΔR*sin(θ j )…(5), in x j Filter tire feature center x mid , and take the tire feature center x mid ±ΔR tolerance target feature points, if the number of target feature points is greater than or equal to c, then the target feature points are tire feature points;

[0030] Among them, b and c are natural numbers, and b and c are adjusted according to radar performance parameters and actual vehicle conditions.

[0031] Optionally, the tire feature center x mid For x j The median, mode or mean of the horizontal distance selected.

[0032] Optionally, after step S8 is completed, the tire feature points are screened out.

[0033] Optionally, step S6 includes:

[0034] Delete the storage point a that first enters the target array n , shift the target array to the right, i.e. a i =a i+1 , where i=1,2…(n-1), target array After that, fill element a1 with the result.

[0035] Compared with existing technologies, the present invention offers the following advantages: The present invention's technical solution effectively eliminates false target points generated by tires, preventing interference with trajectory data processing. Furthermore, the present invention accurately locates tire positions, incorporates tire feature points into functional logic, identifies target types, and initiates early warning. Furthermore, without changing the overall signal processing flow, the present invention effectively utilizes algorithm hardware without significantly increasing computational complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flowchart of an embodiment of the present application.

[0037] Figure 2 A schematic diagram of determining the Doppler search range of a target wheel according to an embodiment of the present application.

[0038] Figure 3 This is a schematic diagram of the constant virtual warning in an embodiment of the present application. DETAILED DESCRIPTION

[0039] The present application will be further described below in conjunction with specific implementation methods.

[0040] The same or similar numbers in the drawings of the embodiments of the present application correspond to the same or similar parts; in the description of the present application, it should be understood that if there are terms such as "upper", "lower", "left", "right", "top", "bottom", "inside", "outside", etc. indicating orientations or positional relationships, they are based on the orientations or positional relationships shown in the drawings, which are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting this patent.

[0041] In addition, if terms such as "first" and "second" are used for descriptive purposes only, they are mainly used to distinguish different devices, elements or components (the specific types and structures may be the same or different), and are not used to indicate or imply the relative importance and quantity of the indicated devices, elements or components, and cannot be understood as indicating or implying relative importance.

[0042] In such Figure 1-3 In the illustrated embodiment, the present application provides a wheel feature recognition method for use in automotive electronic products. The method includes:

[0043] S1, determining the Doppler search range of the target wheel and obtaining the initial signal of the target wheel within the search range; in step S1, determining the Doppler search range of the target wheel includes: determining the number n of sampling points spanned by the wheel in the distance dimension by using the distance resolution, and searching the wheel data using the data of n distance items as the search range; the search range is determined by the following formula: Where 1 is the wheel diameter.

[0044] S2, performing constant virtual warning processing on the Doppler dimension of the distance item of the initial signal to obtain a target array; step S2 includes: performing sliding window selection processing on the initial signal in sequence to select target points, and filtering the target points by peak value to obtain a target array.

[0045] S3, by accumulating the elements of the target array, the target points are obtained; in step S3, the Doppler dimension CFAR is continuously performed on the initial signal until Each element in has a result, and the element results are accumulated to get the total target number m, which is the target number of points on the wheel.

[0046] S4: Determine whether the target number of tire points is greater than or equal to a preset number of tire points. In step S4, the preset number of tire points is a manually set value and is adjusted based on radar performance parameters and actual vehicle conditions. Determining whether the target number of tire points is greater than or equal to the preset number of tire points determines whether to proceed to step S5.

[0047] S5, if based on the target number of points, obtain the wheel feature array; In step S5, based on the target number of points, obtain the wheel feature array, including: obtaining the current distance item number γ, calculating the wheel target position k, the formula is: in, R max is the maximum recognition distance; the distance item of the wheel target position k is set to the tire feature observation center point, the tire feature data K is stored and the target array is cleared.

[0048] S6, otherwise repeat steps S1-S4 to fill the target array iteratively. In step S6, it includes: deleting the storage point a that first enters the target array n , shift the target array to the right, i.e. a i =a i+1 , where i=1,2…(n-1), target array After that, fill element a1 with the result.

[0049] In this embodiment, during the driving process of the car, the speed detection of the target vehicle at close range will cause interference, of which a more serious part is caused by the tires of the target vehicle. The feature recognition of the tires of the present invention is mainly based on the fact that the relative speed of the wheel to the radar at the same distance will produce a micro-Doppler effect, forming many mirror speed points. Based on this feature, a large number of interference points caused by the tires can be eliminated within the alarm range of the close-range BSD angle radar, effectively reducing the probability of inaccurate close-range speed measurement. The wheel feature array is obtained through steps S1-S6, and then the wheel feature array is processed to effectively eliminate false target points generated by the tires, preventing interference with the trajectory during data processing. Moreover, the present application can find the tire position more accurately, and tire feature points can be added to the functional logic to identify the target type and enter the alarm in advance. At the same time, the present application effectively utilizes the algorithm hardware without increasing the amount of calculation without changing the overall signal processing process.

[0050] In some embodiments, in step S1, determining the Doppler dimension search range of the target wheel includes: determining the number n of sampling points spanned by the wheel in the distance dimension by using the distance resolution, and searching the wheel data using the data of the n distance items as the search range; the search range is determined by the following formula: Wherein, l is the wheel diameter. In this embodiment, see Figure 1 , the general wheel diameter is l, and the search range of Doppler dimension CFAR can be determined according to the distance resolution: That is, the number of sampling points spanning the wheel in the distance dimension is determined by resolution as n, and the data for n distance items is used as the wheel data range for the search. l is the wheel diameter, which can be obtained from radar data or confirmed by human input. In this embodiment, l can be the average or mode of the tire diameters.

[0051] In some embodiments, in step S2, it includes: performing sliding window selection processing on the initial signal in sequence to select the target point, and filtering the target point through peak value to obtain the target array. Performing sliding window selection processing on the initial points in sequence to select the target point includes: setting protection units on the adjacent two sides of the object to be inspected, and dividing the selection area on both sides of the protection unit into left window X and right window Y; performing summing and averaging calculations on each point of the left window X and the right window Y, and comparing the amplitude of the object to be inspected with a preset adaptive threshold to obtain the target point. Filtering the target point through peak value to obtain the target array includes: deleting the results of the surrounding points with an amplitude higher than the target point, and storing the remaining target points in the target array Target array Among them, a i The larger the value, the more targets there are at the same distance and different speeds, and the more similar the wheel state is. Figure 3D is the object under inspection, and protection units are located on either side of D. The protection units do not perform constant false alarm processing. The selected areas on the left and right sides of the two protection units are the left window X and the right window Y, respectively. The sum and average calculations for the left window X and the right window Y are performed, and an adaptive threshold is output for comparison with the amplitude of the object under inspection. The preset adaptive threshold can be any one of the following: the average of the left window X and the right window Y, the smaller of the left window X and the right window Y, or the larger of the left window X and the right window Y. The preset adaptive threshold can be selected based on the scenario.

[0052] In some embodiments, in step S5, the wheel feature array is obtained according to the number of target points, including: obtaining the current distance item number γ, and calculating the wheel target position k, which is calculated as follows: in, R max is the maximum recognition distance; the distance term of the wheel target position k is set as the tire feature observation center point, the tire feature data K is stored and the target array is cleared. In this embodiment, when the number of target points m is greater than or equal to the preset number of tire points, then according to the distance term γ at this time, according to the following formula Find the wheel center point, calculate the tire target position k, set the k-point distance item as the tire feature observation center point, store it in the tire feature array K, and clear the target array γ.

[0053] In some embodiments, steps S2-S5 are repeated to perform constant false alarm processing on all initial signals. If the tire characteristic data K is not empty, tire characteristic points are obtained according to the tire characteristic data K. Specifically, the following steps are included:

[0054] S7, the tire characteristic observation center point k of the tire characteristic data K i As the wheel center, calculate the longitudinal distance: y j =k i *ΔR*cos(θ j )…(4), the vertical distance range is or Points h within the range are selected. In step S7, the distance range can be adjusted according to different usage scenarios. If the lateral distance is close, the tire edge angle is too small, resulting in similar distance parameters. Due to the large error in close-range angle measurement, if the radar performance is slightly weaker, the longitudinal distance range in step S7 can be replaced with a distance range that does not include angles. This can more effectively find tire feature points. In step S7, a point set with large velocity dispersion can also be added as tire feature points, which is applicable in scenarios with high point cloud density.

[0055] S8, if point h is greater than b, calculate the horizontal distance and direction by angle: x j =k i *ΔR*stn(θj )…(5), in x j Filter tire feature center x mid , and take the tire feature center x mid ±ΔR tolerance target feature point, if the number of target feature points is greater than or equal to c, then the target feature point is the tire feature point; in step 1, the tire feature center x mid For x j The median, mode or average value of the horizontal distance is selected. In this embodiment, the order is sorted according to the size of the horizontal distance, and the median x is taken. mid The lateral position of the tire is represented by the target point with a ±ΔR tolerance. The lateral distance parameter can be switched to a distance parameter, and the median calculation can be performed. The method for finding the tire feature center can also be adjusted based on actual conditions, such as taking the mode or average. b and c are natural numbers and are adjusted based on radar performance parameters and actual vehicle conditions.

[0056] In some embodiments, after step S8 is completed, the tire feature points are screened out to prevent the tire feature points screened out in step S8 from being substituted into the calculation of the data processing part as false target points.

[0057] In some embodiments, step S6 includes: deleting the storage point a that first enters the target array. n , shift the target array to the right, i.e. a i =a i+1 , where i=1,2…(n-1), target array Then fill element a1 with the result. In this embodiment, the target array of this application is It can be an array with a fixed storage length n. When m is less than the preset number of tire points a, the storage point a that first enters the array is deleted. n , shift the array to the right, i.e. a i =a i+1 , where i = 1, 2…(n-1), after the array is updated, loop step 3 and fill element a1 with the result.

[0058] The technical solution of this application effectively eliminates false target points generated by tires, preventing them from interfering with trajectory data processing. Furthermore, this application can accurately locate tire positions, adding tire feature points to functional logic to identify target types and initiate early alarms. Furthermore, without changing the overall signal processing flow, this application effectively utilizes algorithmic hardware without significantly increasing computational complexity.

[0059] Obviously, the above embodiments of the present application are merely examples for the purpose of clearly illustrating the present application, and are not intended to limit the embodiments of the present application. For those skilled in the art, other variations or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present application shall be included within the scope of protection of the claims of the present application.

Claims

1. A wheel feature recognition method, characterized in that: Applied to automotive electronic products, the method includes: S1, determining the Doppler search range of the target wheel and obtaining the initial signal of the target wheel within the search range; S2, performing constant virtual warning processing on the range-Doppler dimension of the initial signal to obtain a target array; S3, obtaining the target points by accumulating the elements of the target array; S4, determining whether the target number of points is greater than or equal to the preset number of tire points; S5, if based on the number of target points, obtain the wheel feature array, specifically including: obtaining the current distance item number , calculate the wheel target position k, set the distance item of the wheel target position k as the tire feature observation center point, store the tire feature data K and clear the target array; S6, otherwise repeat steps S1-S4 to iterate and fill the target array.

2. A wheel feature recognition method according to claim 1, characterized in that: In step S1, determining the Doppler search range of the target wheel includes: The number of sampling points n spanned by the wheel in the distance dimension is determined by the distance resolution, and the data of n distance items are used as the search range of the wheel data for searching; The search range is determined by the following formula: …(1), where l is the wheel diameter.

3. A wheel feature recognition method according to claim 1, characterized in that: In the step S2, it includes: The initial signal is sequentially subjected to sliding window selection processing to select target points, and the target points are subjected to peak screening to obtain a target array.

4. A wheel feature recognition method according to claim 3, characterized in that: The step of sequentially performing a sliding window selection process on the initial points to select a target point includes: A protection unit is set on two adjacent sides of the inspected object, and the selected area on both sides of the protection unit is divided into a left window X and a right window Y; Perform sum and average calculation on each point of the left window X and the right window Y, compare the amplitude of the object to be inspected with a preset adaptive threshold, and obtain the target point.

5. A wheel feature recognition method according to claim 4, characterized in that: The preset adaptive threshold is any one of an average value of the left window X and the right window Y, a smaller value of the left window X and the right window Y, and a larger value of the left window X and the right window Y.

6. A wheel feature recognition method according to claim 4, characterized in that: The target points are filtered through peak values ​​to obtain a target array, including: Delete the results of the surrounding points whose amplitude is higher than the target point, and store the remaining target points in the target array ; Target array ;in, The larger it is, the more targets there are at the same distance but different speeds, and the more similar it is to the wheel state.

7. A wheel feature recognition method according to claim 6, characterized in that: In step S5, the current distance item number is obtained , calculate the target wheel position k, the formula is: …(2), where …(3), is the maximum recognition distance.

8. A wheel feature recognition method according to claim 7, characterized in that: Repeat steps S2 to S5 to perform constant false alarm processing on all initial signals. If the tire characteristic data K is not empty, obtain tire characteristic points according to the tire characteristic data K.

9. A wheel feature recognition method according to claim 8, characterized in that: The step of obtaining tire feature points according to the tire feature data K specifically includes the following steps: S7, the tire characteristic observation center point of the tire characteristic data K As the wheel center, calculate the longitudinal distance: …(4), the vertical distance range is [ ]or[ ] points h in the filter are selected; S8, if point h is greater than b, calculate the horizontal distance and direction using the angle: …(5), in Tire Feature Center , and take the tire feature center ± The target feature points of the tolerance, if the number of target feature points is greater than or equal to c, then the target feature points are tire feature points; Among them, b and c are natural numbers, and b and c are adjusted according to radar performance parameters and actual vehicle conditions.

10. A wheel feature recognition method according to claim 9, characterized in that: The tire feature center For The median, mode or mean of the horizontal distance selected.

11. A wheel feature recognition method according to claim 9, characterized in that: After step S8 is completed, the tire feature points are screened out.

12. A wheel feature recognition method according to claim 6, characterized in that: In the step S6, it includes: Delete the first storage point that enters the target array , shift the target array to the right, that is ,in , target array After that, fill the element result.

Images